research report on soccer

Evolution of soccer as a research topic

Affiliation.

• 1 James R. Urbaniak, MD Sports Sciences Institute Duke Health, Durham, North Carolina. Electronic address: [email protected].
• PMID: 32599029
• DOI: 10.1016/j.pcad.2020.06.011

Soccer has not only the largest number of worldwide participants, it is also the most studied sport, with nearly 14,000 citations listed on Pubmed and nearly 60% more articles than the next most studied sport. Research about soccer was limited until the late 1970s when exponential growth began; approximately 98% of all soccer-related research publications have occurred since 1980. This vast repository of soccer research shows trends in various major (e.g., 'sex' or 'age group' or 'performance' or 'injury') and specialty (e.g., agility, deceleration, elbow-head impact injuries, behavior) topics. Examining trends of the various topics provides insights into which subjects have come in and out of favor as well as what topics or demographics have been neglected and worthy of inquiry. A further examination can be used by students to learn the most productive researchers, which programs have a strong history of inquiry, and what journals have demonstrated a commitment to publishing research on soccer.

Keywords: Association football; Pubmed; Research history.

Copyright © 2020 Elsevier Inc. All rights reserved.

Publication types

• Age Factors
• Biomedical Research / trends*
• Cardiorespiratory Fitness
• Health Status
• Middle Aged
• Periodicals as Topic / trends*
• Sex Factors
• Soccer / trends*
• Time Factors
• Young Adult

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Data Descriptor
• Open access
• Published: 28 October 2019

A public data set of spatio-temporal match events in soccer competitions

• Luca Pappalardo   ORCID: orcid.org/0000-0002-1547-6007 1 ,
• Paolo Cintia 2 ,
• Alessio Rossi 2 ,
• Emanuele Massucco 3 ,
• Paolo Ferragina 2 ,
• Dino Pedreschi 2 &
• Fosca Giannotti 1  

Scientific Data volume  6 , Article number:  236 ( 2019 ) Cite this article

92k Accesses

83 Citations

391 Altmetric

Metrics details

• Information technology
• Interdisciplinary studies
• Medical research

Soccer analytics is attracting increasing interest in academia and industry, thanks to the availability of sensing technologies that provide high-fidelity data streams for every match. Unfortunately, these detailed data are owned by specialized companies and hence are rarely publicly available for scientific research. To fill this gap, this paper describes the largest open collection of soccer-logs ever released, containing all the spatio-temporal events (passes, shots, fouls, etc.) that occured during each match for an entire season of seven prominent soccer competitions. Each match event contains information about its position, time, outcome, player and characteristics. The nature of team sports like soccer, halfway between the abstraction of a game and the reality of complex social systems, combined with the unique size and composition of this dataset, provide an ideal ground for tackling a wide range of data science problems, including the measurement and evaluation of performance, both at individual and at collective level, and the determinants of success and failure.

Machine-accessible metadata file describing the reported data: https://doi.org/10.6084/m9.figshare.9711164

Similar content being viewed by others

Predicting the in-game status in soccer with machine learning using spatiotemporal player tracking data

SC2EGSet: StarCraft II Esport Replay and Game-state Dataset

Synchronization of passes in event and spatiotemporal soccer data

Background & summary.

Soccer analytics has attracted interest for a long time 1 , 2 . In the early 1950s Charles Reep collected statistics by hand to suggest that “the key to scoring goals is to transfer the ball as quickly as possible from back to front” 3 , thereby indirectly starting the long-ball movement in English football 4 .

Apart from a few sporadic attempts, it is only in recent years that soccer statistics have developed, thanks to sensing technologies that provide high-fidelity data streams extracted from every match. There are three main data sources available 5 : (i) soccer-logs describe the events that occur during a match and are collected through proprietary tagging software 6 , 7 , 8 , 9 ; (ii) video-tracking data describe the trajectories of players during a match and are collected through video recordings 10 , 11 ; (iii) GPS data describe the trajectories of players during training sessions and are collected through GPS devices worn by the players 12 . Despite this wealth of data, we cannot avoid noticing that soccer datasets are rarely available for scientific research. This limits the development of scientific methods for soccer analytics.

In this paper, we describe an open collection of soccer-logs that cover seven prominent male soccer competitions. The collection has been used recently during the Soccer Data Challenge initiative ( https://sobigdata-soccerchallenge.it/ ) and, to the best of our knowledge, it is the largest collection of soccer-logs ever released to the public. Soccer-logs describe match events , each containing information about its type (pass, shot, foul, tackle, etc.), a time-stamp, the player(s), the position on the field and additional information (e.g., pass accuracy). We believe that these data are greatly beneficial to the scientific community because they can contribute to research in several directions, such as the ones we outline below.

Performance analysis

Soccer-logs can be used to design algorithms for relevant problems such as the evaluation of performance and the discovery of tactics 1 , 5 . The problem of performance evaluation 9 , 13 , 14 is crucial for many actors in the sports industry: from broadcasters who want to solicit critical analysis among the fans, to managers who want to monitor the quality of their players and scouts who aim to improve the retrieval of talents. The automatic discovery of tactics 6 , 15 is also becoming a crucial task: while most tactical analyses are currently performed by reviewing video and matches in person, soccer-logs can be used to perform automatic discovery of tactics, simplifying the complex process of match analysis. While different approaches have been proposed in the literature using different datasets to attack these problems, our dataset is much larger and can serve as a common ground to compare and validate different solutions.

Complex systems analysis

Two soccer teams in a match represent a complex system whose global behavior depends in subtle ways on the dynamics of the interactions among the players. Soccer-logs enable the representation of a team as a network , in which nodes represent players and the edges interactions between nodes, usually passes 7 , 14 . While the structure of passing networks is proven to be linked to a team’s strength 7 , 14 , the potential of a multiplex and dynamic representation of networks in soccer has not been much investigated 16 . Soccer-logs allow the definition of different types of interactions between both teammates and opponents by relying on the several event types they encode. Such a richness of information, combined with the dichotomous nature of soccer matches (where collaboration and competition coexist), provides an unprecedented opportunity to investigate novel aspects about the dynamics of complex networks.

Science of success

The availability of a large dataset of sports performance also creates the opportunity to explore the relationship between performance and success, where a team’s success can be intended as its outcome in a competition and the player’s as their popularity or market value. While this relationship has been investigated for individual sports 17 , 18 , apart from a few attempts 19 , 20 there is not much work for soccer, partly due to the absence of publicly available datasets of performance. Our dataset gives the unprecedented opportunity to answer fascinating questions like: ‘What are the tactical patterns of successful teams?’, ‘What are the factors influencing a player’s popularity and market value?’ and ‘To what extent is success predictable from the observable performance?’

The data described in this paper have been collected and provided by Wyscout, a leading company in the soccer industry which connects soccer professionals worldwide, supports more than 50 soccer associations and more than 1,000 professional clubs around the world. The procedure of data collection is performed by expert video analysts (the operators), who are trained and focused on data collection for soccer, through a proprietary software (the tagger). The tagger has been developed and improved over several years and it is constantly updated to always guarantee better and better performance at the highest standards. Based on the tagger and the videos of soccer games, to guarantee the accuracy of data collection, the tagging of events in a match is performed by three operators, one operator per team and one operator acting as responsible supervisor of the output of the whole match. Optionally for near-live data delivery a team of four operators is used, one of them acting to speed up the collection of complex events which need additional and specific attributes or a quick review.

The tagging of a match consists of three main steps.

Step 1: setting formations

At the beginning of the match, an operator sets the teams’ starting formations, the positions of the players on the pitch and their jersey number. The formation of a team consists of the list of players in the starting lineup and the list of players on the bench.

Step 2: event tagging

For each ball touch in the match, the operator selects one player and creates a new event on the timeline. The operator then adds the type (e.g., pass, duel, shot, etc.) and subtype (e.g., a duel can be aerial or ground) of the event by using a special custom keyboard which gives operators the possibility to insert events and data in a streamlined way (Fig.  1a ). The operator finally adds the coordinates on the pitch and all the additional attributes for the event. These can be different depending on the event type: such as pass high/low, foot, dribbling side and so forth (Fig.  1b ). When a player shoots on goal, like in the example of Fig.  1b for player n.6 (Koke), the system asks the operator to fill a shot specific module that collects where the shot ends (on goal, out of goal, on post and exact position).

The process of tagging the soccer events from a match video. ( a ) Screenshot from the tagging software. An action is tagged by an operator via a special custom keyboard, thus creating a new event on the match timeline. ( b ) When the event position on the pitch is set, the shot specific input module appears (top). Event related input modules also appear for setting additional attributes of the occurring event (bottom).

Step 3: quality control

After the tagging, a procedure of quality control for each match is performed, mainly consisting of two different steps. The first step is automatic: an algorithm is used to avoid the majority of the errors made by operators, considerably reducing the margin of error. For example the algorithm matches the events tagged by both operators to crosscheck if they both collected events involving both teams, like duels, with the same positioning and interpretation. Similarly, the algorithm suggests events missed by the operators or searches for impossible combinations of event sequences. The second step of quality control is manual and supervised by quality controllers. It mainly consists of an in-depth check that is carried out once the match is completed. Going through each event of some sample matches, the controller can see and eventually correct any entered parameter. Sample matches for quality control are chosen by another algorithm in order to guarantee a well distributed and statistically meaningful coverage with respect to the kind and amount of analyzed matches.

Data Records

The data sets are released under the CC BY 4.0 License and are publicly available on figshare 21 .

The data refer to season 2017/2018 of five national soccer competitions in Europe: Spanish first division, Italian first division, English first division, German first division, French first division. These competitions are the most important in Europe according to the UEFA country coefficient, which is used to rank the football associations of Europe and thus determine the number of clubs from an association that will participate in the UEFA Champions League and the UEFA Europa League ( https://www.uefa.com/memberassociations/uefarankings/country/#/yr/2019 ). In addition, we provide the data of the World cup 2018 and the European cup 2016, which are competitions for national teams. In total, we provide seven data sets corresponding to information about all competitions, matches, teams, players, events, referees and coaches. Each data set is provided in JSON format (JavaScript Object Notation), an open-standard file format that uses human-readable and machine-processable text to transmit data objects consisting of attribute-value pairs and array data types (or any other serializable value). Table  1 shows the list of competitions we make available with their total number of matches, events and players. The data covers a total of around 1,941 matches, 3,251,294 events and 4,299 players.

Competitions

The competitions data set describes the seven competitions. Each competition is a document consisting of the following fields 21 (see Wyscout documentation for further details at https://apidocs.wyscout.com/ ):

area : denotes the geographic area associated with the league as a sub-document, using the ISO 3166-1 specification;

format : the format of the competition. All competitions for clubs (Spanish first division, Italian first division, English first division, German first division, French first division) have value “Domestic league”. The competitions for national teams (World cup 2018, European cup 2016) have value “International cup”;

name : the official name of the competition (e.g., Spanish first division, Italian first division, World cup 2018, etc.);

type : the typology of the competition. It is “club” for the competitions for clubs (Spanish first division, Italian first division, English first division, German first division, French first division) and “international” for the competitions for national teams (World cup 2018, European cup 2016);

wyId : the unique identifier of the competition, assigned by Wyscout.

Box  1 shows a document in the competitions data set referring to the Italian first division (“name”: “Italian first division”), a competition for clubs (“type”: “club” and “format”: “Domestic League”) held in Italy (see field “area”).

The matches data set describes all the matches we make available. Each match is a document consisting of the following fields 21 (see Wyscout documentation for further details at https://apidocs.wyscout.com/ ):

competitionId: the identifier of the competition to which the match belongs to. It is a integer and refers to the field “wyId” of the competition document;

date and dateutc : the former specifies date and time when the match starts in explicit format (e.g., May 20, 2018 at 8:45:00 PM GMT + 2), the latter contains the same information but in the compact format YYYY-MM-DD hh:mm:ss;

duration : the duration of the match. It can be “Regular” (matches of regular duration of 90 minutes + stoppage time), “ExtraTime” (matches with supplementary times, as it may happen for matches in continental or international competitions), or “Penalities” (matches which end at penalty kicks, as it may happen for continental or international competitions);

gameweek : the week of the league, starting from the beginning of the league;

label : contains the name of the two clubs and the result of the match (e.g., “Lazio - Internazionale, 2–3”);

roundID : indicates the match-day of the competition to which the match belongs to. During a competition for soccer clubs, each of the participating clubs plays against each of the other clubs twice, once at home and once away. The matches are organized in match-days: all the matches in match-day i are played before the matches in match-day i  + 1, even tough some matches may be postponed to facilitate players and clubs participating in Continental or Intercontinental competitions. During a competition for national teams, the “roundID” indicates the stage of the competition (eliminatory round, round of 16, quarter finals, semifinals, final);

seasonId : indicates the season of the match;

status : it can be “Played” (the match has officially started and finished), “Cancelled” (the match has been canceled before it started), “Postponed” (the match has been postponed and no new date and time is available yet) or “Suspended” (the match has been suspended by the referee because of conditions which make it impossible to continue play, such as inclement weather or power failure, and no new date and time is available yet);

venue : the stadium where the match was held (e.g., “Stadio Olimpico”);

winner : the identifier of the team that won the game, or 0 if the match ended with a draw;

wyId : the identifier of the match, assigned by Wyscout;

teamsData : it contains several subfields describing information about each team that is playing that match, such as lineup, bench composition, list of substitutions, coach and scores:

hasFormation : it has value 0 if no formation (lineups and benches) is present, and 1 otherwise;

score : the number of goals scored by the team during the match (not counting penalties);

scoreET : the number of goals scored by the team during the match, including the extra time (not counting penalties);

scoreHT : the number of goals scored by the team during the first half of the match;

scoreP : the total number of goals scored by the team after the penalties;

side : the team side in the match (it can be “home” or “away”);

teamId : the identifier of the team;

coachId : the identifier of the team’s coach;

bench : the list of the team’s players that started the match on the bench and some basic statistics about their performance during the match (goals, own goals, cards);

lineup : the list of the team’s players in the starting lineup and some basic statistics about their performance during the match (goals, own goals, cards);

substitutions : the list of team’s substitutions during the match, describing the players involved and the minute of the substitution.

Box  2 shows a document describing a match between Lazio and Internazionale (“label”: “Lazio - Internazionale, 2–3”) of the Italian first division (“competitionId”: 524), held on May 20th 2018 (see fields “date” and “dateutc”). Box  3 shows the structure of the formation subdocument for one of the teams, which includes the list of players on the bench, the list of players in the starting lineup and the list of substitutions made by the team.

The teams data set describes the clubs or national teams playing in the seven competitions. Each document in this data set consists of the following fields 21 (see Wyscout documentation for further details at https://apidocs.wyscout.com/ ):

city : the city where the team is located. For national teams it is the capital of the country;

name : the common name of the team;

area : information about the geographic area associated with the team;

wyId : the identifier of the team, assigned by Wyscout;

officialName : the official name of the team (e.g., Juventus FC);

type : the type of the team. It is “club” for teams in the competitions for clubs (Spanish first division, Italian first division, English first division, German first division, French first division.) and “national” for the teams in international competitions (World cup 2018, European cup 2016);

Box  4 shows a document describing team Juventus (“name”: “Juventus”) which is located in Turin (“city”: “Torino”) in Italy (see field “area”).

The players data set describes all players in the seven competitions 21 (see Wyscout documentation for further details at https://apidocs.wyscout.com/ ). Each document in this data set consists of the following fields:

birthArea : geographic information about the player’s birth area;

birthDate : the birth date of the player, in the format “YYYY-MM-DD”;

currentNationalTeamId : the identifier of the national team where the players currently plays;

currentTeamId : the identifier of the team the player plays for. The identifier refers to the field “wyId” in a team document;

firstName : the first name of the player;

lastName : the last name of the player;

foot : the preferred foot of the player;

height : the height of the player (in centimeters);

middleName : the middle name (if any) of the player;

passportArea : the geographic area associated with the player’s current passport;

role : the main role of the player. It is a subdocument containing the role’s name and two abbreviations of it;

shortName2 : the short name of the player;

weight : the weight of the player (in kilograms);

wyId : the identifier of the player, assigned by Wyscout.

Box  5 shows a document describing player Lionel Andres Messi Cuccittini (“shortName2”: “L. Messi”), who was born in Argentina (see field “birthArea”) and has the Spanish passport (see “passportArea”). From the document we observe that Messi’s preferred foot is the left foot (“foot”: “left”), his height and weight are 170 centimeters (“height”: 170) and 72 kilograms (“weight”: 72) respectively, he preferably plays as a forward (see field “role”) and he was born in 1987 (“birthDate”: “1987-06-24”).

The events data set describes all the events that occur during each match 21 (see Wyscout documentation for further details at https://apidocs.wyscout.com/ ). Each event document contains the following information:

eventId : the identifier of the event’s type. Each eventId is associated with an event name (see next point);

eventName : the name of the event’s type. There are seven types of events (see Table  2 ): pass, foul, shot, duel, free kick, offside and touch;

subEventId : the identifier of the subevent’s type. Each subEventId is associated with a subevent name (see next point);

subEventName : the name of the subevent’s type. Each event type is associated with a different set of subevent types (see Table  2 );

tags : a list of event tags, each describing additional information about the event (e.g., accurate). Each event type is associated with a different set of tags (see Table  2 ). The Wyscout documentation provides a mapping of the tag identifiers to the corresponding names and descriptions ( https://apidocs.wyscout.com/ );

eventSec : the time when the event occurs (in seconds since the beginning of the current half of the match);

id : a unique identifier of the event;

matchId : the identifier of the match the event refers to. The identifier refers to the field “wyId” in a match document;

matchPeriod : the period of the match. It can be “1H” (first half of the match), “2H” (second half of the match), “E1” (first extra time), “E2” (second extra time) or “P” (penalties time);

playerId : the identifier of the player who generated the event. The identifier refers to the field “wyId” in a player document;

positions : the origin and destination positions associated with the event. Each position is a pair of coordinates ( x , y ). The x and y coordinates are always in the range [0, 100] and indicate the percentage of the field from the perspective of the attacking team. In particular, the value of the x coordinate indicates the event’s nearness (in percentage) to the opponent’s goal, while the value of the y coordinates indicates the event’s nearness (in percentage) to the right side of the field;

teamId : the identifier of the player’s team. The identifier refers to the field “wyId” in a team document.

Box  6 shows an example of pass event (“eventId”: 8, “eventName”: “Pass”) generated by player 3344 (“playerId”: 3344) of team 3161 (“teamId”: 3161) in match 2576335 (“matchId”: 2576335) at second 2.41 of the first half of the match (“eventSec”: 2.4175, “matchPeriod”: “1H”). The pass started at position (49, 50) of the field and ended at position (38, 58) of the field (see field “positions”). Moreover, the pass was accurate as indicated by the presence of tag 1801 (field “tags”).

The coaches data set describes all coaches of the clubs and the national teams of the seven competitions we make available 21 (see Wyscout documentation for further details at https://apidocs.wyscout.com/ ). It consists of the following fields:

wyId : the identifier of the coach, assigned by Wyscout.

shortName : the short name of the coach;

firstName : the first name of the coach;

middleName : the middle name (if any) of the coach;

lastName : the last name of the coach;

birthDate : the birth date of the coach, in the format “YYYY-MM-DD”;

birthArea : geographic information about the coach’s birth area;

passportArea : the geographic area associated with the coach’s current passport;

currentTeamId : the identifier of the coach’s team. The identifier refers to the field “wyId” in a team document.

Box  7 shows a document describing coach Maurizio Sarri (“shortName”: “M. Sarri”), who was born in Italy (see field “birthArea”), has the Italian passport (see “passportArea”) and he was born in 1959 (“birthDate”: “1959-01-10”).

The referees data set describes all referees in the national and international competitions we make available 21 (see Wyscout documentation for further details at https://apidocs.wyscout.com/ ). It consists of the following fields:

wyId : the identifier of the referee, assigned by Wyscout.

shortName : the short name of the referee;

firstName : the first name of the referee;

middleName : the middle name (if any) of the referee;

lastName : the last name of the referee;

birthDate : the birth date of the referee, in the format “YYYY-MM-DD”;

birthArea : geographic information about the referee’s birth area;

passportArea : the geographic area associated with the referee’s current passport;

Box  8 shows a document describing referee William Collum (“shortName”: “W. Collum”), who was born in Scotland (see field “birthArea”), has a Scottish passport (see “passportArea”) and was born in 1979 (“birthDate”: “1979-01-18”).

Box 1 Example of document in the competitions data set describing the Italian first division.

Box 2 Example of a document in the matches data set describing a match between Lazio and Internazionale.

Box 3 Example of team document describing the club Juventus FC.

Box 4 Information about a team in the teams data set.

Box 5 Information about a player contained into the players data set.

Box 6 Information about a pass event contained into the events data set.

Box 7 Information about a coach in the coaches data set.

Box 8 Information about a referee contained into the referee data set.

Technical Validation

In general, based on the events data set, a soccer match consists of an average of 1,682 ± 101 events (Fig.  2b ), with an inter-time between two consecutive events of 3.59 ± 7.42 seconds. There are on average 59 ± 29 events observed for a player in a match, one every 78.78 ± 105.64 seconds, confirming that soccer players are typically in ball possession for less than two minutes 22 . Passes are the most frequent events, accounting for around 50% of the total events (Fig.  2a ). Duels (e.g., tackles and dribbles) are the second most frequent events (≈28%), while shots account for about 1.5% of the total events. The goals scored, the most important events in soccer since they determine a match outcome, are the rarest ones accounting for less than 1% of the total number of events. We provide an example of all the events (1,620) observed for the match “Lazio - Internazionale” of the Italian first division (May 20, 2018), plotted on the position of the field where they have occurred (Fig.  2c ).

Statistics of the events data set. ( a ) Frequency of events per type. ( b ) Distribution of the number of events in soccer matches. ( c ) Events produced by the two teams in the match Lazio (cyan points) vs. Internazionale (black squares). The events are plotted on the position of the field where they occurred.

Spatial dimension

By looking at the position of the field where the events occur, we can investigate interesting aspects of a soccer match, such as the spatial distribution of players and events. For example the kernel density plot in Fig.  3a shows that passes are distributed mostly in the center of the field, where actually most of the match takes place. As one could expect, we observe differences in the spatial distribution of events when we select the players by their role: while the events of forwards are observed mainly in the opponent’s half of the field (Fig.  3h ), the events of defenders are observed mostly in the own half and on the sides of the field (Fig.  3g ). Similarly, as expected the spatial distribution of events change with their type: attacking events (e.g., shots) are mostly observed close to the opponent’s goal (Fig.  3b ), while defensive events (e.g., clearances) are mostly observed close to the team’s own goal (Fig.  3f ). The spatial dimension of match events can provide us with information about a player’s behavior during a match, giving for example the possibility to determine a player’s profile from his average position during a match 13 .

Distribution of positions per event type. ( a–f ) Kernel density plots showing the distribution of the events’ positions during match. The darker is the green, the higher is the number of events in a specific field zone. ( g–i ) Distribution of the passes’ position during a match for each player’s role. The darker is the color, the higher is the number of passes in a specific field zone.

Temporal dimension

By looking at when the events occur during a game, we can investigate interesting dynamics of teams and players. For example, Fig.  4 shows that goals are scored more frequently in the second half of the match 23 , 24 , mirroring several of the possible factors that could affect scoring, such as a decrease of attention by the defenders towards the end of the match due to a loss of stamina, or a more offensive attitude of the opponents who try to win or equalize the match. Similarly, we observe that the frequency of other rare events like yellow and red cards is the highest in the recovery time. This aspect could highlight the presence of a bias by the referees who are less prone to award a card in the beginning of a match (as suggested in 25 ), a reduction of stamina or an increment of aggression of players at the end of the match.

In-match evolution of the number of events. Number of events (i.e., goals on the top plot, yellow cards in the middle plot and the red cards in the bottom plot) that occur in all the matches in the data set, with time windows of 5 minutes.

Another aspect that can be investigated by combining the spatial and the temporal dimensions of soccer-logs are the so-called invasion index , a measure of how close to the opponent’s goal a team plays during a match (i.e., its dangerousness), and acceleration index , a measure of how fast a team reaches the closest position to the opponent’s goal 26 . By exploiting the spatial and temporal dimension of soccer-logs, the invasion index can be computed on each possession phase, which is defined as a sequence of events on the ball made by a team before the opponents gain the possession. To compute the invasion index of a possession phase we compute: (i) for each event in the possession phase, the probability of scoring from the position where the event occurs (defined as the fraction of goals that have been scored from that position); (ii) we take the highest of these probabilities. A team’s overall invasion index during a match is simply the average invasion index across its possession phases. Figure  5 shows the invasion and acceleration index of the teams throughout the match Roma - Fiorentina (0–2), played on April 7, 2018. We observe that Fiorentina has on average a higher invasion index than Roma (0.27 ± 0.33 and 0.23 ± 0.31, respectively).

Invasion index and acceleration index for a game in the match data set. Bold lines represent the rolling mean of, respectively, invasion index ( a ) and acceleration index ( b ), while thin lines represent the individual values computed for each possession phase of each team. Purple vertical lines refer to the two goal scored by Fiorentina during the match, while the red vertical line indicates the half time of the match.

A team’s average acceleration index is another measure of its playing efficacy during a match. The acceleration index of a team’s possession phase is computed as the ratio between its invasion index and the square of the time between the first event and most dangerous event of the possession phase. A team’s average acceleration index during a match is the average acceleration index across its possession phases. Similarly to the invasion index, Fiorentina has a higher average acceleration than Roma (Roma: 0.06 ± 0.16, Fiorentina: 0.07 ± 0.15).

Both the invasion and the acceleration indices show that Fiorentina (the winner of the match) was more dangerous during the match, staying closer to the opponent’s goal and reaching dangerous zones faster than Roma.

Team analysis

Soccer-logs enable the analysis of the interactions between players through the reconstruction of a team’s passing network 7 , 14 , a representation of the movements of the ball between teammates during a match. A passing network allows identifying the key players in the team, i.e., the ones having more connections to the teammates or a high passing activity 27 , 28 . Figure  6 shows two examples of a team passing network for the match Napoli - Juventus (Italian first division). Although Napoli engaged in more passes than Juventus (666 vs. 332), the two passing networks show similar average weighted out-degrees (1.01 ± 0.93% and 1.10 ± 0.84%, respectively). However, Juventus’ playing style resulted in a higher connectivity 29 , defined as the network’s second smallest eigenvalue (i.e., a root of the characteristic equation of a matrix). This value indicates the robustness of a team, i.e., the strength of the links between its players. As a matter of fact, large values of connectivity between teammates are associated with a better overall team performance.

Representation of the player passing networks of the match Napoli-Juventus. Nodes represent players, edges represent passes between players. The size of the nodes reflects the number of ingoing and outgoing passes (i.e. node’s degree), while the size of the edges is proportional to the number of passes between the players.

The reconstruction of passing networks from soccer-logs enables several performance analyses 7 . For example, by using the passing network and the players’ position during a pass it is possible to identify the most efficient tactical patterns across teams 30 , 31 .

Player analysis

Soccer-logs can be used to compare the performance of players and track their evolution in time. As an example, we compare three forwards with different characteristics – L. Messi (FC Barcelona), C. Ronaldo (Juventus FC) and M. Salah (Liverpool). We observe that L. Messi has the highest passing activity: while he produces 49 ± 19 passes per match on average, C. Ronaldo and M. Salah produce 26 ± 6 and 25 ± 9 passes per matches, respectively. Additionally, we observe that L. Messi engages in more duels per match (25 ± 8) than C. Ronaldo and M. Salah (15 ± 5 and 21 ± 7 duels per match). The data we release to the public also enable the computation of several performance metrics, such as Flow Centrality 14 and PlayeRank 13 . A player’s flow centrality in a match is defined as his betweenness centrality in the passing network 14 . Figure  7a shows the distribution of flow centrality of L. Messi, C. Ronaldo and M. Salah for the matches in season 2017/2018. L. Messi results in a higher flow centrality (0.10 ± 0.01) than C. Ronaldo and M. Salah (0.09 ± 0.01 and 0.09 ± 0.01, respectively).

Distribution of flow centrality and PlayeRank score for three top players. ( a ) Distribution of the flow centrality of L. Messi (red line), C. Ronaldo (blue line) e M. Salah (black line) during the soccer season 2017/2018. ( b ) Performance quality calculated as the PlayeRank score of L. Messi (red line), C. Ronaldo (blue line), and M. Salah (black line).

The performance quality of the players during the season can be assessed using PlayeRank, a data-driven framework that offers a principled multi-dimensional and role-aware evaluation of the soccer players’ performance quality in a match or in a series of matches 13 . Figure  7b shows that the three aforementioned players have different performance trends during the season. M. Salah obtained his best performance in the first part of the season, then decreasing during the course of the season. In contrast, L. Messi significantly increases his performance quality throughout the season while C. Ronaldo, who was not playing the first part of the season due to an injury, has on average a performance quality slightly higher than Salah but lower than Messi. We can conclude that, according to two measures computed on soccer-logs, Messi performs the best both in terms of passing centrality and performance quality.

Code availability

The code to reproduce the plots in the paper is available upon request by writing at [email protected] or [email protected].

Bornn, L., Cervone, L. D. & Fernández, J. Soccer analytics: Unravelling the complexity of “the beautiful game”. Significance 15 , 26–29 (2018).

Article   Google Scholar  

Anderson, C. & Sally, D. The Numbers Game: Why Everything You Know About Football is Wrong. Penguin Books (2013)

Reep, C. & Benjamin, B. Skill and Chance in Association Football. Journal of the Royal Statistical Society 131 , 581–585 (1968).

Sykes J. & Paine N. How One Man’s Bad Math Helped Ruin Decades Of English Soccer. Five Thirty Eight (2016)

Gudmundsson, J., Butte, A. J. & Horton, M. Spatio-Temporal Analysis of Team Sports. ACM Computing Surveys 50 (2), 22:1-22:34 (2017).

Decroos, T., Van Haaren, J. & Davis, J. Automatic Discovery of Tactics in Spatio-Temporal Soccer Match Data. Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (KDD ’18) , 223–232 (2018).

Cintia, P., Pappalardo, L., Pedreschi, D., Giannotti, F. & Malvaldi, M., The harsh rule of the goals: Data-driven performance indicators for football teams. Proceedings of the 2015 IEEE International Conference on Data Science and Advanced Analytics (DSAA) , 1–10 (2015).

Cintia, P., Rinzivillo, S. & Pappalardo, L. Network-based Measures for Predicting the Outcomes of Football Games. Proceedings of the 2nd Workshop on Machine Learning and Data Mining for Sports Analytics (MLSA) , 46–54 (2015).

Brooks, J., Kerr, M. & Guttag, J. Developing a Data-Driven Player Ranking in Soccer Using Predictive Model Weights. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD ’16) , 49–55 (2016).

Bornn, L. & Fernendez, J. Wide Open Spaces: A statistical technique for measuring space creation in professional soccer. MIT Sloan Sports Analytics Conference 2018 (2018).

Wei, X., Sha, L., Lucey, P., Morgan, S. & Sridharan, S. Large-Scale Analysis of Formations in Soccer. Proceedings of the 2013 International Conference on Digital Image Computing: Techniques and Applications (DICTA) , 1–8 (2013).

Rossi, A. et al . Effective injury forecasting in soccer with GPS training data and machine learning. PloS One 13 (7), 1–15 (2018).

Pappalardo, L. et al . PlayeRank: data-driven performance evaluation and player ranking in soccer via a machine learning approach. ACM Transactions on Intelligent Systems and Technology (TIST), 10 (5), 59:1-59:27 (2018).

Duch, J., Waitzman, J. S. & Amaral, L. A. N. Quantifying the Performance of Individual Players in a Team Activity. PlosOne 5 (6), 1–7 (2010)

Bialkowski, A. et al . Large-Scale Analysis of Soccer Matches Using Spatiotemporal Tracking Data. Proceedings of the 2014 IEEE International Conference on Data Mining (ICDM) , 725–730 (2014).

Buldú, J. M. et al . Using Network Science to Analyse Football Passing Networks: Dynamics, Space, Time, and the Multilayer Nature of the Game. Frontiers in Psychology 9 , 1664–1078 (2018).

Yucesoy, B. & Barabasi, A.-L. Untangling performance from success. EPJ Data Science 5 (1), 1–17 (2016).

Cintia, P., Pappalardo, L. & Pedreschi, D. “Engine Matters”: A First Large Scale Data Driven Study on Cyclists’ Performance. Proceedings of the 13th IEEE International Conference on Data Mining Workshops , 147–153 (2013).

Pappalardo, L. & Cintia, P. Quantifying the relation between performance and success in soccer. Advances in Complex Systems 21 (3) (2017).

Pappalardo, L., Cintia, P., Pedreschi, D., Giannotti, F. & Barabasi, A.-L. Human Perception of Performance. Preprint at, http://arxiv.org/abs/1712.02224 (2017).

Pappalardo, L. & Massucco, E. Soccer match event dataset. figshare , https://doi.org/10.6084/m9.figshare.c.4415000.v2 (2019).

Link, D. & Hoerning, M. Individual ball possession in soccer. PLoS One 12 (7), 1–15 (2017).

Armatas, V., Yiannakos, A., Papadopoulou, S. & Skoufas, D. Evaluation of goals scored in top ranking soccer matches: Greek “Superleague” 2006-07. Serbian Journal of Sports Sciences 3 (1), 39–43 (2009).

Alberti, G., Iaia, F. M., Arcelli, E., Cavaggioni, L. & Rampinini, E. Goal scoring patterns in major European soccer leagues. Sport Sciences for Health 9 (3), 151–153 (2013).

Unkelbach, C. & Memmert, D. Game Management, Context Effects, and Calibration: The Case of Yellow Cards in Soccer Journal of Sport and Exercise Psychology 1 , 95–109 (2008).

Link, D. & Weber, H. Using individual ball possession as a performance indicator in soccer. Workshop on Large-Scale Sports Analytics (2015)

Gama, J. et al . Network analysis and intra-team activity in attacking phases of professional football. International Journal of Performance Analysis in Sport  14 (3), 692–708 (2014).

Passos, P. et al . Networks as a novel tool for studying team ball sports as complex social systems.  Journal of Science and Medicine in Sport 14 (2), 170–176 (2011).

Article   CAS   Google Scholar  

Clemente, F. M., Martins, F. M. L., Kalamares, D., Wong, D. P. & Mendes, R. S. General network analysis of national soccer teams in FIFA World Cup 2014. International Journal of Performance Analysis in Sport 15 , 80–96 (2015).

Wang, Q., Zhu, H., Hu, W., Shen, Z. & Yao, Y. Discerning tactical patterns for professional soccer teams: an enhanced topic model with applications. 21 th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (2015).

Rein, R. & Memmert, D. Big data and tactical analysis in elite soccer: future challenges and opportunities for sports science. SpringerPlus 5 (1) (2016).

Download references

Acknowledgements

This work has been partially funded by EU project SoBigData RI, grant #654024.

Author information

Authors and affiliations.

ISTI-CNR, Pisa, Italy

Luca Pappalardo & Fosca Giannotti

Department of Computer Science, University of Pisa, Pisa, Italy

Paolo Cintia, Alessio Rossi, Paolo Ferragina & Dino Pedreschi

Wyscout, Chiavari, Italy

Emanuele Massucco

You can also search for this author in PubMed   Google Scholar

Contributions

L.P. designed data sets, processed data, made plots and wrote the paper. P.C. designed data sets and processed data. A.R. processed data, made plots and wrote the paper. E.M. designed data sets and wrote the paper. P.F. wrote the paper. D.P. and F.G. provided funding.

Corresponding author

Correspondence to Luca Pappalardo .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ .

The Creative Commons Public Domain Dedication waiver http://creativecommons.org/publicdomain/zero/1.0/ applies to the metadata files associated with this article.

Reprints and permissions

About this article

Cite this article.

Pappalardo, L., Cintia, P., Rossi, A. et al. A public data set of spatio-temporal match events in soccer competitions. Sci Data 6 , 236 (2019). https://doi.org/10.1038/s41597-019-0247-7

Download citation

Received : 28 February 2019

Accepted : 08 August 2019

Published : 28 October 2019

DOI : https://doi.org/10.1038/s41597-019-0247-7

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing: AI and Robotics newsletter — what matters in AI and robotics research, free to your inbox weekly.

• Open access
• Published: 24 August 2016

Big data and tactical analysis in elite soccer: future challenges and opportunities for sports science

• Robert Rein 1 &
• Daniel Memmert 1  

SpringerPlus volume  5 , Article number:  1410 ( 2016 ) Cite this article

58k Accesses

208 Citations

56 Altmetric

Metrics details

Until recently tactical analysis in elite soccer were based on observational data using variables which discard most contextual information. Analyses of team tactics require however detailed data from various sources including technical skill, individual physiological performance, and team formations among others to represent the complex processes underlying team tactical behavior. Accordingly, little is known about how these different factors influence team tactical behavior in elite soccer. In parts, this has also been due to the lack of available data. Increasingly however, detailed game logs obtained through next-generation tracking technologies in addition to physiological training data collected through novel miniature sensor technologies have become available for research. This leads however to the opposite problem where the shear amount of data becomes an obstacle in itself as methodological guidelines as well as theoretical modelling of tactical decision making in team sports is lacking. The present paper discusses how big data and modern machine learning technologies may help to address these issues and aid in developing a theoretical model for tactical decision making in team sports. As experience from medical applications show, significant organizational obstacles regarding data governance and access to technologies must be overcome first. The present work discusses these issues with respect to tactical analyses in elite soccer and propose a technological stack which aims to introduce big data technologies into elite soccer research. The proposed approach could also serve as a guideline for other sports science domains as increasing data size is becoming a wide-spread phenomenon.

Tactics are a central component for success in modern elite soccer. Yet until recently, there have been few detailed scientific investigations of team tactics. One reason in this regard has been the lack of available, relevant data. With the development of advanced tracking technologies this situation has changed recently. Instead, now the amount of available data is becoming increasingly difficult to manage. In the present article we discuss how recent developments of big data technologies from industrial data analytics domains address these problems. Further, the present work provide an overview how big data technologies may provide new opportunities to study tactical behavior in elite soccer and what future challengers lie ahead.

Soccer tactics background

According to the Oxford dictionary, tactics describe “an action or strategy carefully planned to achieve a specific end”. Regarding competitive soccer, naturally the aim the end of the activity is to win the game. Choosing an appropriate tactic is therefore crucial for every pre-game preparation (Carling et al. 2005b ; Kannekens et al. 2011 ; Sampaio and Macas 2012 ; Yiannakos and Armatas 2006 ). Regarding the definition of tactics Gréhaigne and Godbout ( 1995 ) introduced a distinction between the strategy and tactics. Here, the team strategy describes the decisions made before the game with respect to how the team wants to play whereas the tactic is the result of the ongoing interactions between the two opposing teams. This approach seems somewhat counter to the basic definition of the term tactics provide above. Furthermore, it is not clear how these two concepts can be clearly delineated from each other as the real-time interactions between the players will be conditioned by the a priori strategy. Following a classical practitioner’s approaches the tactic specifies how a team manages space, time, and individual actions to win a game (Fradua et al. 2013 ; Garganta 2009 ). In this context, space specifies for example were on the pitch a certain actions takes places or which area a team wants to occupy during the attack and the defense. Time in contrast describes variables like frequency of events and durations (ball possession) or how quick actions are being initiated. For example, a team could decide to have a slow buildup during attack initiate in the defense third where individual players hold the ball for longer times whereas in the attacking third only fast on-touch pass sequences are preferred. Finally, individual actions specify the type of actions which are being performed, for example turnovers, crosses and passes (Garganta 2009 ). This classification can be further hierarchically organized along the number of participating players into individual tactics, group tactics, team tactics, and match tactics which is also a scheme commonly referred to by soccer practitioners (Bisanz and Gerisch 1980 , p.201; Carling et al. 2005a ). Individual tactics describe all one-on-one events during offensive and defensive play with and without the ball. For example, the way the ball carrier is approached by a defender can be considered as part of the individual tactic. For example, the defender could immediately attack the ball carrier and put him under pressured or the defender could use a more passive approaches focusing mainly on blocking passing channels. Group tactics describe the cooperation between sub groups within a team for example the defensive block during an offside trap. Team tactics describe preferred offensive and defense team formations (e.g. 4-4-2) and the positioning of the formation on the pitch (Grunz et al. 2012 ). Finally, game tactics describe the team’s playing philosophy such as counter-attack or ball possession play. A recent study investigated for example ball possession regain in the German Bundesliga where the results showed that more successful teams were faster to regain ball possession after losing possession (Shafizadehkenari et al. 2014 ; Vogelbein et al. 2014 ). In summary, soccer tactics describe the microscopic and macroscopic organizational principles of the players on the pitch spanning from individual to group decision making processes.

To ensure successful execution at all tactical levels, a coach has to take into account the status of the team, the status of the opposition, as well as external factors like playing at home or even the weather (Gréhaigne and Godbout 1995 ; Lago 2009 ; Mackenzie and Cushion 2013 ; Sarmento et al. 2014 ) (compare Fig.  1 ). Therefore, in the following tactics refers to both the a priori decisions as well as the real-time adaptations during a game. As the two competing teams try to out-smart each other, the tactics are not constant but should be adapted according to the interactions between and within the two teams (Balagué and Torrents 2005 ; Garganta 2009 ; Grehaigne et al. 1997 ; Gréhaigne and Godbout 2014 ). For example, a player substitution by the opposition team may introduce a change in playing tactics which the coach may have to respond to be changing his teams’ tactics. Team tactics are therefore governed by a complex process resulting from a network of inter-dependent parameters (Kempe et al. 2014 ). Although the scheme presented above follow a hierarchical pattern the flow information in reality does go in both directions. Tactics at a higher level condition the tactics at the lower level and vice versa success of individual actions equally conditions success at a higher level (Araújo et al. 2006 ; Sampaio and Macas 2012 ). Thus, tactics can be interpreted as complex structure of composed of a new of interwoven dependencies. Accordingly, tactical analysis should reflect this complexity.

Overview of factors influencing tactics in soccer

Over the years tactical decisions, like preferred playing formations or game tactics, have increased in complexity and coaches’ tactical abilities are under constant public scrutiny. Until very recently this stood somewhat in contrast to the amount of scientific investigations studying tactical decisions in elite soccer (Carling et al. 2005c ; Garganta 2009 ; Sampaio and Macas 2012 ; Sarmento et al. 2014 ). The reason for this somewhat surprising fact may have been the lack of accessible and/or reliable data required for tactical analysis (Rampinini et al. 2007 ). The present gold standard to assess tactical behavior and team performance in general in elite soccer is commonly based on individual game observations (Dutt-Mazumder et al. 2011 ; Mackenzie and Cushion 2013 ). A domain expert (coach, scout) observes a game and rates the team tactics according to his personal experiences. Although usually a specific coding manual is used a general consensus regarding relevant variables is currently missing (James 2006 ; Sarmento et al. 2014 ) and data often lack objectivity and reliability (James et al. 2002 ). Furthermore, as game interactions are highly dynamic and contextual circumstances change continually it is under debate to what extent reliable measures are attainable in general (Lames and McGarry 2007 ). In addition, detailed game analyses based on observational approaches are highly time-consuming which limited their application in the past (Carling et al. 2008 ; James 2006 ). Consequently, demand for more quantitative oriented (automatic) approaches to analyze tactical behavior in elite soccer is increasing (Beetz et al. 2005 ; Carling et al. 2014 ; Lucey et al. 2013a , b ; Wang et al. 2015 ). Thus, whereas the processes underlying tactics in elite soccer have increased over the years the scientific approaches have not quite evolved with the same speed.

In this regard, fine-grained global reporting of game event statistics for commercial audiences has seen a tremendous rise in recent years and detailed game data are routinely reported (Baca 2008 ; Baca et al. 2004 ; Sarmento et al. 2014 ). The reason for this increased availability of game data is largely due to progress made in player tracking technologies (Baca 2008 ; Carling et al. 2008 ; Castellano et al. 2014 ; D’Orazio and Leo 2010 ; Lu et al. 2013 ). Recently FIFA the governing body for international competitive soccer decided to allow the usage of wireless sensors technologies to track player positions and physiological parameters during competitions (di Salvo and Modonutti 2009 ). This will further increase the availability of detailed performance data from elite soccer. Thereby this has been a results of today’s common practices among professional teams to already collect physiological and tracking data during training and friendly matches to manage the training process (Bush et al. 2015 ; Carling et al. 2008 ; Ehrmann et al. 2016 ; Goncalves et al. 2014 ; Ingebrigtsen et al. 2015 ). At present, several different tracking systems are available in the market including vision based systems, Global Positioning Systems (GPS), and radio wave based tracking systems (Leser et al. 2011 ). Although data quality and reliability used to be a problem, in recent years the systems have matured to such an extent that the data is now of sufficient quality to satisfy scientific standards. Several recent overviews addressing the advantages and disadvantages between the different available systems are available in the literature (Barris and Button 2008 ; Buchheit et al. 2014 ; Carling et al. 2008 ; Castellano et al. 2014 ; D’Orazio and Leo 2010 ; Harley et al. 2011 ; Valter et al. 2006 ). Thus modern tracking data allows the analysis of technical, tactical and physical demands in elite soccer.

In general, a trend seems to emerge where analyses of soccer games in public media outlets are also becoming increasingly data aware. One example in this regard is the increasing number of free internet blogs reporting detailed game analyses. Using observational techniques from TV game broadcasts data as well as publicly available internet soccer databases these blogs provide novel approaches to data driven performance analysis in soccer much in the same spirits as the sabermetrics community has for American baseball during the late 90’s (Lewis 2004 ). Recently, investigations have emerged which used sentiment analysis from twitter feeds to identify for example high impact events during games (Buntain 2014 ; Yu and Wang 2015 ) and to predict game outcomes (Godin et al. 2014 ). In this regard, quantified-self initiatives may also provide future opportunities to generate valuable data for scientific investigations (Appelboom et al. 2014 ; Shull et al. 2014 ). In summary, lack of reliable data to perform tactical analysis in elite soccer is becoming less of a problem and novel data sources are continually being discovered and developed.

Analysis of soccer tactics

Traditionally, one area which has produced a wealth of studies investigating soccer performance is with respect to the physiological demands in competitive soccer (Carling et al. 2008 ; Mohr et al. 2005 ). However, until recently few connections between physiological demands and tactical behavior in elite soccer have been made (Bloomfield et al. 2007 ; Drust et al. 2007 ; Moura et al. 2012 ). As was made clear in the introduction, the success for a tactics depends on the abilities of the individual players to actually implement the required actions. Obviously this requires that the players fulfill the necessary physiological requirements, for example, when playing a ball possession type of play (da Mota et al. 2016 ). Rampinini et al. ( 2007 ) investigated the total running distances and the time spent different running speed categories (standing to sprinting). The results showed a significant influence of the level of the opponents and the playing position (compare also Goncalves et al. 2014 ). Bush et al. ( 2015 ) investigated the changes in physiological performance variables in the English Premier League across several seasons and results indicated significant increases in passing event rates associated with changes in team tactics (Bush et al. 2015 ). Carling ( 2011 ) investigated the influence of opposition formations on physiological and skill-related performance variables and found for example increased running distances when playing against a 4-2-3-1 formation compared to a 4-4-2 formation (Carling 2011 ). Sampaio et al. ( 2014 ) investigated the influence of time unbalance and game pace on physiological demands during a 5-a-side small sided game were one player was dropped in either side to create an inferiority or an superiority condition. The results suggested an effect of team unbalance on the time spent in different hear rate zones suggesting that the inferior team had to work harder (Sampaio et al. 2014 ). In summary, these results indicate that tactical behavior and physiological variables are linked but more in-depth analyses are missing. Accordingly, at present it is unclear how to combine information about player’s physiology from training and competition with team tactics (Castellano et al. 2014 ) and no connections between individual technical performance and team tactics have been made so far (Hughes and Bartlett 2002 ).

Traditionally, tactics analyses relied on notational analysis approaches based on average statistics and tallies (Hughes and Bartlett 2002 ). Indicators include for example passing variables (Hughes and Franks 2005 ; Liu et al. 2015 ), ball possession (Collet 2013 ; Lago 2009 ), ball recovery (Vogelbein et al. 2014 ), or playing style (Tenga et al. 2010a , b ). The main limitation of the traditional notational approach is that almost all contextual information is discarded, these measures have shown weak explanatory power with limited adoption by practitioners (Glazier 2015 ; Hughes and Bartlett 2002 ; Mackenzie and Cushion 2013 ; Nevill et al. 2008 ; Sarmento et al. 2014 ; Tenga et al. 2010a , b ). To circumvent this problem increasingly multi-variate approaches are being used to retain contextual information (Fernandez-Navarro et al. 2016 ; Kempe et al. 2014 ). Almeida et al. ( 2016 ) investigated the effect of different scoring modes on ball-recovery type and location, playing configuration and defensive state in youth players. The results showed that more ball recoveries were made when a central goal was used and that most recoveries were a result of set-play in the defensive third of the pitch. Younger players also produced more elongated shapes in the playing direction whereas the older teams produced more flattened shapes with larger spread in the direction orthogonal to the playing direction (Almeida et al. 2016 ). Tenga et al. ( 2010a , b ) investigated the effects of a ten different variables on score-box possession based on video data from 163 matches from the Norwegian men’s professional league in 2004. The results showed that the odds ratio for producing a score-box possession increased when the attacking team had a long possession, started their attack from the final third, or used penetrative passes against a balanced defense. However, counterattack, possession starting in the final third, long possession, long pass, and penetrative passes showed increased odds ratios against an imbalanced defense. Recently, Fernandez-Navarro et al. ( 2016 ) used 19 performance indicators to identify different playing styles. The results showed that several factors like possession directness which correlated with ball possession, sideway passes, and passes from the defensive third into the attacking third were important to identify playing styles (Fernandez-Navarro et al. 2016 ).

One approach which is increasingly being used to study team tactics is the team centroid method (Folgado et al. 2014 ; Frencken et al. 2011 , 2012 ; Yue et al. 2008 ). Here the behavior of the team centroid, the geometric center of the positions of all players from a team, is used to analyze the behavior of the whole team. Results from this line of research indicate a strong coupling between team centroids during game play (Frencken et al. 2011 ), changes of inter-centroid distances due to pitch size variations (Duarte et al. 2012a , b ; Frencken et al. 2013 ), and key game events like goal shots are accompanied by increased inter-team coupling variability (Frencken et al. 2012 ). More recently, investigation of centroid behavior has been further extended by calculating the Approximate Entropy (ApEn) (Pincus and Goldberger 1994 ), a non-linear time-series measurement techniques, to quantify the regularity in time-series data (Aguiar et al. 2015 ; Goncalves et al. 2014 ; Sampaio and Macas 2012 ). Results using ApEn analysis suggest increased centroid behavior regularity after tactical training in novice players (Duarte et al. 2012a , b ; Sampaio and Macas 2012 ). Goncalves et al. ( 2014 ) investigated the coordination during on 11-a-side game between and within the defenders, mid-fielders, and attacker subgroups using ApEn. The results showed that players movements were more regular with respect to the centroid of their respective groups compared to the other groups. Sampaio et al. ( 2014 ) further showed that during an inferiority situation during a 5-a-side small sided game the regularity of the distance to the team centroid increased. Goncalves et al. ( 2016 ) investigated the influence of numerical imbalances between attacking and defending team in small sided games in professional and amateur players. Player numbers varied between 4 versus 3, 4 versus 5, and 4 versus 7. The results showed that in experts an increase in the number of opponents increased the regularity in team behavior with respect to the opponents. Although the application of ApEn is becoming more prominent, it still remains to be shown what this measure really represents as the regularity behavior of team centroids in itself represent a highly abstract description of team behavior. Nevertheless, team centroid measures increasingly are being used to capture team behavior and many interesting applications have been reported in the literature in recent years.

Another more recent group of approach to study team tactics focuses on the control of space. On such approach uses for example the team surface area as calculated from the convex hull which encloses all players from one team (Frencken et al. 2011 ; Moura et al. 2012 , 2013 ). Results from this line of research indicates that greater surface areas are covered by the attacking compared to the defensive teams (Frencken et al. 2011 ; Moura et al. 2012 ). Similar, more experienced players also cover a greater area compared to less experienced players (Duarte et al. 2012a , b ; Olthof et al. 2015 ). Fradua et al. ( 2013 ) investigated the individual player area during 11-a-side matches by calculating the largest rectangle enclosing all field players divided by the number of players. The results showed that individual playing areas become smaller when the ball moved into the central pitch area. Another approach uses Voronoi-diagrams to investigate space control (Nakanishi et al. 2008 ). Here the controlled space is determined using the location and distances between individual players to determine the controlled space. Results using Voronoi-diagrams show similar results compared to the team surface area approach (Fonseca et al. 2012 ; Fujimura and Sugihara 2005 ; Gudmundsson and Wolle 2014 ; Kim 2004 ; Taki and Hasegawa 2000 ) Finally, another approach is based on the determination of numerical superiority in a particular pitch area (Silva et al. 2014 ). Together these results indicate that space control is a central aspect of soccer tactics and further highlight the interactive nature underlying soccer games (Duarte et al. 2013 ; Garganta 2009 ; Grehaigne et al. 1997 ; Tenga et al. 2010a , b ).

Another emerging analysis approach to study team tactics studies investigates team passing behavior using network approaches (Watts and Strogatz 1998 ). The basic rationale of this approach is to model the players of a team as nodes and the passes occurring between them as weighted vertices where the number of passes between two players determine the weights (Duarte et al. 2012a , b ; Passos et al. 2011 ). This representation of team passing behavior allows to easily identify key players within in a team as they display more connection to other vertices accompanied by greater vertex weights (Gama et al. 2014 ; Passos et al. 2011 ). Recent network analyses which included next to the player information also pass position information were able to predict game outcomes and the final ranking of the top teams using a K-Nearest Neighbor classifier (Cintia et al. 2015 ). Similar, Wang et al. ( 2015 ) used Bayesian latent model approach applied to passing network and passing position information from 241 games from the Spanish First (2013–2014). The obtained model was able to automatically identify different tactical patterns across teams. By combining the obtained tactical information with attacking success the authors were further able to show which specific tactical patterns were more efficient across teams. By investigating the contributions by the individual players to each tactical pattern the authors were further able to determine individual contributions by the players to each tactical pattern (Wang et al. 2015 ). Together these results suggest that players interactions mediated through passing behavior in combination with spatial information provides an interesting new approaches to analyze tactical behavior in elite soccer thereby providing much more information compared to traditional notational analysis approaches.

Increasingly tactical decision making in elite soccer is also investigated using machine learning (ML) algorithms based on game position data (Bialkowski et al. 2014a , b ; Fernando et al. 2015 ; Xinyu et al. 2013 ). Machine learning algorithms allow to identify specific data patterns in large datasets by building an a priori unknown model from the data (Haykin 2009 ; Jordan and Mitchell 2015 ; Waljee and Higgins 2010 ). Although this approach has been discussed in sports research for some time (Bartlett 2004 ; Borrie et al. 2002 ; Nevill et al. 2008 ) only recently successful applications become more prevalent (Bartlett 2004 ; Lucey et al. 2013a , b ). For example, application of an expectation maximization algorithm with position data from an entire English Premier League season allowed the automatic identification of team formations (Bialkowski et al. 2014a , b ; Lucey et al. 2013a , b ). The results further showed that teams used more defensive formations during away games (Bialkowski et al. 2014a , b ). The authors used a two-step algorithm where the formations were identified only after each player was assigned a specific role. This approach allowed the authors to circumvent the problem that the player’ roles are not constant throughout the game but change according to the context which precludes the possibility to simply use the id of each individual player to identify team formations (Bialkowski et al. 2014a ; Lucey et al. 2013a , b ). Knauf et al. ( 2016 ) used a spatio-temporal kernel algorithm to cluster trajectories which allowed automatic differentiated game initiation and scoring opportunities from position data. Pairwise similarities between trajectories during attacking phases were compared using a specific metric and subsequently a clustering algorithms grouped the trajectories into clusters. Again, one of the underlying features of the algorithm used by the authors is that the comparison between trajectories is invariant to permutations between players (Knauf et al. 2016 ). Using spatial tracking data, Kihwan et al. ( 2010 ) applied a temporal kernel method to predict the location of the ball on the pitch. By calculating a flow-field from the running directions of the players the authors were able to determine convergence points of flow-field which predicted future positions of the ball with good agreement (Kihwan et al. 2010 ). Hirano and Tsumoto ( 2005 ) used a multiscale comparison technique with combined event data type and event location data to automatically identify reoccurring attacking sequences leading to a goal. The multiscale comparison technique allowed to compare event sequences of varying length with each other. For example, in the spatial-kernel method this problem has been resolved by time-normalizing the data (Knauf et al. 2016 ). Similar, Fernando et al. ( 2015 ) were able to differentiate attacking plays across teams using cluster analysis of game sequences (compare also Xinyu et al. 2013 ). Recently, Montoliu et al. ( 2015 ) applied a Bag-of-Words algorithm to coding soccer game video snippets followed by a Random Forest classifier to identify game play patterns. The authors divided the pitch into ten areas and calculated the optical flow representing the moving direction of players during short video sequences extracted from two complete soccer game recording. Thus, the application relied on the pre-segmentation of the raw video data by experts (Montoliu et al. 2015 ).

A second group of ML approaches featuring prominent in the soccer literature uses neural network modeling (compare Dutt-Mazumder et al. 2011 for a comprehensive overview). Here, in particular Kohonen Feature Maps (KFM) have been used to study tactical patterns (Barton et al. 2006 ; Bauer and Schöllhorn 1997 ; Dutt-Mazumder et al. 2011 ; Kohonen 1990 , 2001 ; Lees and Barton 2003 ). For example, Grunz et al. ( 2012 ) used a Hierarchically Dynamically Controlled Network KFM (Perl 2002 , 2004 ; Perl and Weber 2004 ) to automatically identify team formations (Grunz et al. 2012 ; Kempe et al. 2015 ; Memmert and Perl 2009 ). In summary, numerous machine learning studies of have used soccer data to study tactical decision making with little guidance for non-experts. Common to these approaches is that mostly a certain facet of team tactics, predominantly team formations, was investigated. Accordingly, information how to combine the information across tactical domains (Fig.  1 ) is lacking currently (Garganta 2009 ; Glazier 2015 ). For example it is not clear how group formations interact with the individual technical and tactical skills of players. As it is clear that different tactical positions within a team have different physiological demands there has been no research addressing how this information can be used in combination with tactical formations used by the attacking and defensive teams (Carling et al. 2008 ). Furthermore, with respect to the tactics hierarchy introduced in the introduction (compare also Fig.  1 ) the presented approaches work at the team tactics level. Accordingly, how team formations influence group tactics of subgroups and individual tactics has not been investigated so far. An interesting side-note of the presented studies is the fact that most ML soccer analyses are performed by computer scientist research group with little apparent involvement by sports scientists.

This short overview shows that although many interesting analyses are available what is lacking is a conceptual connection between them. Accordingly, it appears that the main obstacle to study team tactics stems from the lack of a theoretical model (Garganta 2009 ; Glazier 2015 ; Mackenzie and Cushion 2013 ). One model which has been repeatedly proposed in the literature is based on a Dynamic system theoretical framework (Duarte et al. 2012a , b ; Duarte et al. 2013 ; Garganta 2009 ; McGarry et al. 2002 ; Reed and Hughes 2006 ; Ric et al. 2016 ). However, although this approach merits great potential, at present already the basic definition of a relevant phase space is lacking. In the dynamic systems theoretical approaches, the phase space constitutes a key concept which describes a theoretical abstractions describing mathematically a space where the system resides in and which enable to capture the dynamics of the system in a meaningful manner (Nevill et al. 2008 ; Vogel 1999 ). Current suggestions regarding appropriate phase space variables in team game vary widely (Duarte et al. 2012a , b ; Gréhaigne 2011 ; Grehaigne et al. 1997 ; Gréhaigne and Godbout 2014 ; Lames and McGarry 2007 ). In this regard, a common approach for example is to use the relative phase as a measure to capture coordination phenomena between players (Duarte et al. 2013 ; Goncalves et al. 2014 ; Sampaio and Macas 2012 ). Relative phase approaches stem from the domain of physical dynamical systems were oscillators typically constitute the building blocks of the systems (Pikovsky et al. 2003 ). Accordingly, the question of whether an oscillator assumption is justified to model team games is an open question at present. Modeling efforts of soccer games as a dynamic system which go beyond a purely phenomenological description are therefore not available at present.

The lack of a higher-order description about soccer team dynamics also prevents the current analytical approaches from making a real impact with practitioners (Carling et al. 2008 ; Lames and McGarry 2007 ; Nevill et al. 2008 ). One of the challenges for tactical match analysis in elite soccer will be to work towards an explanatory theoretical model which is able to integrate information from various domains including tactics, physiology, and motor skills (Garganta 2009 ; Sarmento et al. 2014 ) (compare Fig.  1 ). In this regard, new approaches in Artificial Intelligence (AI) research (Bishop 2013 ; Gibney 2016 ; Jones 2014 ; LeCun et al. 2015 ) may provide promising avenues towards the development of a theoretical model of tactical decision making in elite soccer. In particular, so-called deep learning networks are becoming increasingly powerful in modeling domains previously considered computational intractable (Hinton and Salakhutdinov 2006 ; LeCun et al. 2015 ; Xue-wen and Xiaotong 2014 ). However, these approaches rely on large training datasets to determine network parameters (Jones 2014 ; Xue-wen and Xiaotong 2014 ), which at present have not been used in tactical analyses in soccer. In this regard, recent machine learning models using neural networks have been extended such to allow to incorporate a priori information into the models (Bishop 2013 ). This might be of great relevance to develop novel approach to model team tactical behaviors as for example insights gained from the studies summarized above might be used to constrain network modeling efforts and at the same time allowing the connection between physiological, tactical and skill related information. Accordingly, modern algorithm from AI might prove highly useful for tactical analysis in elite soccer and fulfill previous proposals (Dutt-Mazumder et al. 2011 ).

Big data and soccer tactics

A potential solution with respect to model building and the combination various data sources might present itself through the recent rise of big data technologies which has been already suggested as shaping the future of performance analysis in elite soccer (Cassimally 2012 ; Kasabian 2014 ; Lohr 2012 ; Medeiros 2014 ; Norton 2014 ). As the phenomenon of big data is relatively recent first a definition of the relevant concepts will be provided. Surprisingly, no universally agreed definition of big data is available and big data is rather described by its characteristics (Baro et al. 2015 ; Noor et al. 2015 ; Romanillos et al. 2016 ). Accordingly, big data is characterized using the so-called three V’s: (1) Volume, (2) Variety and (3) Velocity (Noor et al. 2015 ; Xue-wen and Xiaotong 2014 ). Volume describes the magnitude of the data, Variety refers to the heterogeneity of data, and Velocity characterizes the data production rate (Noor et al. 2015 ). With respect to tactical analytics in soccer these concept can be mapped in the following way: (1) Volume refers to the size of datasets in soccer. For example, a current dataset for positional data typically encoded using Extensible Markup Language (XML) ranges between 86 and 300 megabytes (mb). Thus, storing position, event and video data from a single complete Bundesliga season results in 400 gigabytes of tracking data. Accordingly the data volume increases with the addition of other sources including for example physiological or event data. By itself this is far from the petabyte data sizes commonly associated with big data (Pääkkönen and Pakkala 2015 ), yet the main problem is to provide structured access to the data. Common solutions using Excel sheets do not scale well with these data. Big data technologies in contrast provide specific solutions for storing such data sets and make them accessible through specific user interfaces and application programming interfaces (API). (2) Variety refers to different data formats and data sources. Variety can be further distinguished into: (a) structured, (b) semi-structured, and (c) unstructured data. Structured data has a clearly predefined schema describing the data. Structured data allows simple navigation and searching through the data where a relational database system is the canonical example. In contrast, unstructured data lacks a definite schema with video data and text messages being typical examples. Accordingly, semi-structured data falls in between these two extremes and consists of data which lacks a pre-defined structure but may has a variable schema which is often part of the data itself (Sint et al. 2009 ). Current XML data types used for tracking data are examples in this regard (IPTC 2001 ). Thus, in soccer data variety refers to position, video, fitness, training, skill performance, and notational meta-data next to health records and crowd data from blogs. As data access and data processing patterns vary across data types, big data technologies provide specific solutions to combine the information distributed across such datasets. (3) Velocity describes the speed with which novel data is being generated. In soccer, the velocity varies widely between real-time streams from physiological and positional data to delayed data from notational analysis during training and competition. Big data technologies specifically address how to process and store high velocity data. In summary, all three key concepts characterizing big data are highly relevant with respect to tactical analysis in elite soccer and big data technological stacks provide specific solutions to address each of these areas.

A candidate big data soccer technological stack for soccer tactics analyses should be organized along several levels (compare Fig.  2 ). First, the necessary infrastructure to collect the data is required spanning physiological and tracking data in addition to video and observational data. Second, a storage system is required allowing efficient data storage and access. Finally, a processing pipeline has to be established to extract relevant information from the data and to subsequently merge the information to build an explanatory and/or predictive model (Coutts 2014 ). For all these processing levels reporting and visualization capabilities are needed to monitor the different processing steps and communicate the results. Unfortunately, there is no one-to-one mapping between these different components and available technologies. However an in-depth discussion of specific technological solutions is beyond the scope of the present article and more specialized literature is referred to (Noor et al. 2015 ; Pääkkönen and Pakkala 2015 ; Sitto and Presser 2015 ).

Big data technological stack for tactical analysis in elite soccer

Yet, what immediately becomes clear from Fig.  2 is that a significant amount of expertise is needed in order to establish such a system. One area which is facing similar challenges in this respect is the medical health sector (Noor et al. 2015 ; Toga et al. 2015 ; Zhang et al. 2015 ). In the medical area a so-called personalized (stratified) medicine is increasingly seen is a key are of research to improve current practices (Hood et al. 2015 ; Kostkova et al. 2016 ; Zhang et al. 2015 ). Thereby, for personalized medicine to become realizable big data technologies are needed. One key problem in this area is how data is stored and shared across institutions. At present health data is collected and held by government, commercial and public research institutions. This leads to sever limitations with respect to access and data sharing possibilities across these entities due to privacy and security issues (Costa 2014 ; Kong and Xiao 2015 ; Kostkova et al. 2016 ; Toga and Dinov 2015 ). This also applies to soccer data where data is collected by commercial institutions, private clubs, and public research institutions. Accordingly, privacy issues have to be addressed as for example detailed profiles about individual players might have significant career implications and professional soccer teams may be reluctant to share data and possibly forfeit competitive advantages. Thus, data governance issues must be resolved before big data approaches may become viable for soccer research potentially. In the medical sector varies solutions are being investigated including standardized open privacy protection mechanisms which encrypts individual data items (Kong and Xiao 2015 ). Nevertheless, even when access is made available, researchers face the problem that data processing is highly complex and not manageable using common processing pipelines. Experiences from the biomedical sectors shows that in particular smaller research groups lack the required expertise and funding to build the required processing and analysis infrastructures (Bishop 2013 ; Goecks et al. 2010 ; Lynch 2008 ; Marx 2013 ; Noor et al. 2015 ; Sitto and Presser 2015 ). At present, it is also not clear how to ensure that technologies and procedures are made available to researchers lacking the required computer science expertise to build data pipelines of their own. This is already a problem with respect to many of the ML techniques described above.

As computational approaches increasingly become more complex reproducibility issue will also become more important as the development of novel algorithmic approach will become the focus of future publication results (Mesirov 2010 ). In this regard, efforts from biomedical research like the Galaxy project (Goecks et al. 2010 ) may provide a model solution for future big data technologies in sports sciences. The Galaxy project is developed through a collaborative effort across several universities and provides a web-based solution to perform genomic research using big data technologies (Goecks et al. 2010 ; Levine and Hullett 2002 ; Ohmann et al. 2015 ). The project aims to provide a standardized way for researchers to access complex processing algorithms which makes it possible for non-expert users to apply cutting edge analysis technologies to their data (Goecks et al. 2010 ). The system includes a sophisticated documentation solution which allows the storage and presentation of analysis results and documents at the same time the complete processing pipeline ensuring reproducibility of the research results (Goecks et al. 2010 ). The framework was build to be extensible and allows the inclusion of additional procedures through public repositories efforts (Blankenberg et al. 2014 ). This approach may be a model for sports sciences to address not only big data approaches for soccer tactics but more general analysis and data processing problems in other domains as well. Inevitable this will lead to increased collaborative efforts between sports and computer scientists as the sports science community at present lacks the required computational background.

In conclusion, exciting times are emerging for team sports performance analysis as more and more data is going to become available allowing more refined investigations. The adaption of big data technologies for soccer research may therefore provide solutions to some of the key issues outline above. Thus, by providing novel methods to analyze the data and a more comprehensive theoretical model and understanding of tactical team performance in elite soccer may be within reach. This implies however, that future soccer research will have to embrace a stronger multi-disciplinary approach. Performance analysts, exercise scientists, biomechanists as well as practitioners will have to work together to make sense of these complex data sets. As has been pointed out, most of the machine learning approaches presented were performed by computer science research groups. Accordingly, future collaborations between computer and sports scientists may hold the key to apply these complex approaches in a more relevant manner. In turn, relying increasingly on more complex data analysis techniques will also pose new challenges for future sports scientists. Therefore, university curricula will have be augmented to ensure that future students receive the required background training to be able to not only use these techniques but to have at least some understanding of their theoretical and computational underpinnings. The introduction of big data technologies will also require a discussions within the research community of how to share data and techniques across research teams. To make the new insights relevant for practice a tight interchange with practitioners is required. Finally, taking a broader view on the issue of big data and sports science the proposed model for tactical analyses in elite soccer might also prove beneficial for other sports science domains where data sizes are bound to increase as well and accordingly similar problems will surface.

Aguiar M, Goncalves B, Botelho G, Lemmink K, Sampaio J (2015) Footballers’ movement behaviour during 2-, 3-, 4- and 5-a-side small-sided games. J Sports Sci 33(12):1259–1266. doi: 10.1080/02640414.2015.1022571

Article   Google Scholar  

Almeida CH, Duarte R, Volossovitch A, Ferreira AP (2016) Scoring mode and age-related effects on youth soccer teams’ defensive performance during small-sided games. J Sports Sci 34(14):1355–1362. doi: 10.1080/02640414.2016.1150602

Appelboom G, LoPresti M, Reginster JY, Sander Connolly E, Dumont EP (2014) The quantified patient: a patient participatory culture. Curr Med Res Opin 30(12):2585–2587. doi: 10.1185/03007995.2014.954032

Araújo D, Davids K, Hristovski R (2006) The ecological dynamics of decision making in sport. Psychol Sport Exerc 7(6):653–676

Baca A (2008) Tracking motion in sport—trends and limitations. Paper presented at the 9th Australasian conference on mathematics and computers in sport, Math Sport (ANZIAM)

Baca A, Baron R, Leser R, Kain H (2004) A process oriented approach for match analysis in table tennis. In: Lees A, Kahn JF, Maynard IW (eds) Science and racket sports III. Routledge, Abingdon, pp 214–219

Google Scholar  

Balagué N, Torrents C (2005) Thinking before computing: changing perspectives in sport performance. Int J Comput Sci Sport 4:5–13

Baro E, Degoul S, Beuscart R, Chazard E (2015) Toward a literature-driven definition of big data in healthcare. Biomed Res Int 2015:639021. doi: 10.1155/2015/639021

Barris S, Button C (2008) A review of vision-based motion analysis in sport. Sports Med 38(12):1025–1043. doi: 10.2165/00007256-200838120-00006

Bartlett R (2004) Artificial intelligence in technique analysis—past, present and future. Int J Perf Anal Sport 4(2):4–19

Barton G, Lees A, Lisboa PJG, Attfield S (2006) Visualisation of gait data with Kohonen self-organising neural maps. Gait Posture 24:46–53

Bauer HU, Schöllhorn W (1997) Self-organizing maps for the analysis of complex movement patterns. Neural Process Lett 5(3):193–199

Beetz M, Kirchlechner B, Lames M (2005) Computerized real-time analysis of football games. IEEE Pervasive Comput 4(3):33–39. doi: 10.1109/MPRV.2005.53

Bialkowski A, Lucey P, Carr P, Yue Y, Matthews I (2014a) Win at Home and Draw Away: automatic formation analysis highlighting the differences in home and away team behaviors MIT Sloan Sports Analytics Conference. Boston

Bialkowski A, Lucey P, Carr P, Yue Y, Sridharan S, Matthews I (2014b) Large-scale analysis of soccer matches using spatiotemporal tracking data. In: 2014 IEEE international conference on paper presented at the data mining (ICDM). 14–17 Dec 2014

Bisanz G, Gerisch G (1980) Fußball: Training, Technik, Taktik. Rororo, Hamburg

Bishop CM (2013) Model-based machine learning. Philos Trans A Math Phys Eng Sci 371(1984):20120222. doi: 10.1098/rsta.2012.0222

Blankenberg D, Von Kuster G, Bouvier E, Baker D, Afgan E, Stoler N, Nekrutenko A (2014) Dissemination of scientific software with Galaxy ToolShed. Genome Biol 15(2):403. doi: 10.1186/gb4161

Bloomfield J, Polman R, O’Donoghue P (2007) Physical demands of different positions in FA Premier League soccer. J Sports Sci Med 6(1):63–70

Borrie A, Jonsson GK, Magnusson MS (2002) Temporal pattern analysis and its applicability in sport: an explanation and exemplar data. J Sports Sci 20(10):845–852. doi: 10.1080/026404102320675675

Buchheit M, Allen A, Poon TK, Modonutti M, Gregson W, Di Salvo V (2014) Integrating different tracking systems in football: multiple camera semi-automatic system, local position measurement and GPS technologies. J Sports Sci 32(20):1844–1857. doi: 10.1080/02640414.2014.942687

Buntain C (2014) Language-agnostic event detection across sports from twitter and using temporal features. Paper presented at the workshop on large-scale sports analytics (KDD 2014), New York, USA

Bush M, Barnes C, Archer DT, Hogg B, Bradley PS (2015) Evolution of match performance parameters for various playing positions in the English Premier League. Hum Mov Sci 39:1–11. doi: 10.1016/j.humov.2014.10.003

Carling C (2011) Influence of opposition team formation on physical and skill-related performance in a professional soccer team. Eur J Sport Sc 11(3):155–164. doi: 10.1080/17461391.2010.499972

Carling C, Williams AM, Reilly T (2005a) From technical and tactical performance analysis to training drills Handbook of soccer match analysis: a systematic approach to improving performance. Routledge, London, pp 129–147

Carling C, Williams AM, Reilly T (2005b) Handbook of soccer match analysis. Routledge, London

Book   Google Scholar  

Carling C, Williams AM, Reilly T (2005c) What match analysis tells us about succeessful strategy and tactics in soccer Handbook of soccer match analysis: a systematic approach to improving performance. Routledge, London, pp 108–128

Carling C, Bloomfield J, Nelsen L, Reilly T (2008) The role of motion analysis in elite soccer: contemporary performance measurement techniques and work rate data. Sports Med 38(10):839–862

Carling C, Wright C, Nelson LJ, Bradley PS (2014) Comment on ‘performance analysis in football: a critical review and implications for future research’. J Sports Sci 32(1):2–7. doi: 10.1080/02640414.2013.807352

Cassimally KA (2012) Soccer’s big data revolution. http://www.nature.com/scitable/blog/labcoat-life/soccers_big_data_revolution

Castellano J, Alvarez-Pastor D, Bradley PS (2014) Evaluation of research using computerised tracking systems (Amisco and Prozone) to analyse physical performance in elite soccer: a systematic review. Sports Med 44(5):701–712. doi: 10.1007/s40279-014-0144-3

Cintia P, Pappalardo L, Pedreschi D, Giannotti F, Malvaldi M (2015) The harsh rule of the goals: data-driven performance indicators for football teams. In: IEEE international conference on paper presented at the data science and advanced analytics (DSAA), 2015. 36678 2015. 19–21 Oct 2015

Collet C (2013) The possession game? A comparative analysis of ball retention and team success in European and international football, 2007–2010. J Sports Sci 31(2):123–136. doi: 10.1080/02640414.2012.727455

Costa FF (2014) Big data in biomedicine. Drug Discov Today 19(4):433–440. doi: 10.1016/j.drudis.2013.10.012

Coutts AJ (2014) Evolution of football match analysis research. J Sports Sci 32(20):1829–1830. doi: 10.1080/02640414.2014.985450

D’Orazio T, Leo M (2010) A review of vision-based systems for soccer video analysis. Pattern Recogn 43(8):2911–2926

da Mota GR, Thiengo CR, Gimenes SV, Bradley PS (2016) The effects of ball possession status on physical and technical indicators during the 2014 FIFA World Cup Finals. J Sports Sci 34(6):493–500. doi: 10.1080/02640414.2015.1114660

di Salvo V, Modonutti M (2009) Integration of different technology systems for the development of football training. J Sports Sci Med S11:3

Drust B, Atkinson G, Reilly T (2007) Future perspectives in the evaluation of the physiological demands of soccer. Sports Med 37(9):783–805

Duarte R, Araujo D, Correia V, Davids K (2012a) Sports teams as superorganisms: implications of sociobiological models of behaviour for research and practice in team sports performance analysis. Sports Med 42(8):633–642. doi: 10.2165/11632450-000000000-00000

Duarte R, Araujo D, Freire L, Folgado H, Fernandes O, Davids K (2012b) Intra- and inter-group coordination patterns reveal collective behaviors of football players near the scoring zone. Hum Mov Sci 31(6):1639–1651. doi: 10.1016/j.humov.2012.03.001

Duarte R, Araujo D, Correia V, Davids K, Marques P, Richardson MJ (2013) Competing together: assessing the dynamics of team-team and player-team synchrony in professional association football. Hum Mov Sci 32(4):555–566. doi: 10.1016/j.humov.2013.01.011

Dutt-Mazumder A, Button C, Robins A, Bartlett R (2011) Neural network modelling and dynamical system theory: are they relevant to study the governing dynamics of association football players? Sports Med 41(12):1003–1017. doi: 10.2165/11593950-000000000-00000

Ehrmann FE, Duncan CS, Sindhusake D, Franzsen WN, Greene DA (2016) GPS and injury prevention in professional soccer. J Strength Condit Res 30(2):360–367. doi: 10.1519/JSC.0000000000001093

Fernandez-Navarro J, Fradua L, Zubillaga A, Ford PR, McRobert AP (2016) Attacking and defensive styles of play in soccer: analysis of Spanish and English elite teams. J Sports Sci. doi: 10.1080/02640414.2016.1169309

Fernando T, Wei X, Fookes C, Sridharan S, Lucey P (2015) Discovering methods of scoring in soccer using tracking data. Paper presented at the Large-Scale Sports Analytics, Sidney

Folgado H, Lemmink KA, Frencken W, Sampaio J (2014) Length, width and centroid distance as measures of teams tactical performance in youth football. Eur J Sport Sc 14(Suppl 1):S487–S492. doi: 10.1080/17461391.2012.730060

Fonseca S, Milho J, Travassos B, Araujo D (2012) Spatial dynamics of team sports exposed by Voronoi diagrams. Hum Mov Sci 31(6):1652–1659. doi: 10.1016/j.humov.2012.04.006

Fradua L, Zubillaga A, Caro O, Ivan Fernandez-Garcia A, Ruiz-Ruiz C, Tenga A (2013) Designing small-sided games for training tactical aspects in soccer: extrapolating pitch sizes from full-size professional matches. J Sports Sci 31(6):573–581. doi: 10.1080/02640414.2012.746722

Frencken W, Lemmink K, Delleman N, Visscher C (2011) Oscillations of centroid position and surface area of soccer teams in small-sided games. Eur J Sport Sci 11(4):215–223. doi: 10.1080/17461391.2010.499967

Frencken W, Poel H, Visscher C, Lemmink K (2012) Variability of inter-team distances associated with match events in elite-standard soccer. J Sports Sci 30(12):1207–1213. doi: 10.1080/02640414.2012.703783

Frencken W, Plaats J, Visscher C, Lemmink K (2013) Size matters: pitch dimensions constrain interactive team behaviour in soccer. J Syst Sci Complex 26(1):85–93. doi: 10.1007/s11424-013-2284-1

Fujimura A, Sugihara K (2005) Geometric analysis and quantitative evaluation of sport teamwork. Syst Comput Jpn 36(6):49–58. doi: 10.1002/scj.20254

Gama J, Passos P, Davids K, Relvas H, Ribeiro J, Vaz V, Dias G (2014) Network analysis and intra-team activity in attacking phases of professional football. Int J Perform Anal Sport 14(3):692–708

Garganta J (2009) Trends of tactical performance analysis in team sports: bridging the gap between research, training and competition. Rev Port Cien Desp 9(1):81–89

Gibney E (2016) Google AI algorithm masters ancient game of Go. Nature 529(7587):445–446. doi: 10.1038/529445a

Glazier PS (2015) Towards a grand unified theory of sports performance. Hum Mov Sci. doi: 10.1016/j.humov.2015.08.001

Godin F, Zuallaert J, Verndersmissen B, De Neve W, Van der Waller R (2014) Beating the bookmakers: leveraging statistics and Twitter microposts for predicting soccer results. Paper presented at the Workshop on Large-Scale Sports Analytics (KDD 2014), New York, USA

Goecks J, Nekrutenko A, Taylor J, Galaxy T (2010) Galaxy: a comprehensive approach for supporting accessible, reproducible, and transparent computational research in the life sciences. Genome Biol 11(8):R86. doi: 10.1186/gb-2010-11-8-r86

Goncalves B, Figueira BE, Macas V, Sampaio J (2014) Effect of player position on movement behaviour, physical and physiological performances during an 11-a-side football game. J Sports Sci 32(2):191–199. doi: 10.1080/02640414.2013.816761

Goncalves B, Marcelino R, Torres-Ronda L, Torrents C, Sampaio J (2016) Effects of emphasising opposition and cooperation on collective movement behaviour during football small-sided games. J Sports Sci 34(14):1346–1354. doi: 10.1080/02640414.2016.1143111

Gréhaigne J-F (2011) Jean-paul sartre and team dynamics in collective sport. Sport Ethics Philos 5(1):34–45. doi: 10.1080/17511321.2010.536956

Gréhaigne J-F, Godbout P (1995) Tactical knowledge in team sports from a constructivist and cognitivist perspective. Quest 47(4):490–505. doi: 10.1080/00336297.1995.10484171

Gréhaigne J-F, Godbout P (2014) Dynamic systems theory and team sport coaching. Quest 66(1):96–116. doi: 10.1080/00336297.2013.814577

Grehaigne J-F, Bouthier D, David B (1997) Dynamic-system analysis of opponent relationships in collective actions in soccer. J Sports Sci 15(2):137–149. doi: 10.1080/026404197367416

Grunz A, Memmert D, Perl J (2012) Tactical pattern recognition in soccer games by means of special self-organizing maps. Hum Mov Sci 31(2):334–343. doi: 10.1016/j.humov.2011.02.008

Gudmundsson J, Wolle T (2014) Football analysis using spatio-temporal tools. Comput Environ Urban Syst 47:16–27

Harley JA, Lovell RJ, Barnes CA, Portas MD, Weston M (2011) The interchangeability of global positioning system and semiautomated video-based performance data during elite soccer match play. J Strength Cond Res 25(8):2334–2336. doi: 10.1519/JSC.0b013e3181f0a88f

Haykin S (2009) Neural networks and learning machines, 3rd edn. Pearson, Upper Saddle River

Hinton GE, Salakhutdinov RR (2006) Reducing the dimensionality of data with neural networks. Science 313(5786):504–507. doi: 10.1126/science.1127647

Hirano S, Tsumoto S (2005, 6–9 Nov. 2005) Grouping of soccer game records by multiscale comparison technique and rough clustering. Fifth international conference on paper presented at the hybrid intelligent systems, 2005. HIS ‘05

Hood L, Lovejoy JC, Price ND (2015) Integrating big data and actionable health coaching to optimize wellness. BMC Med 13:4. doi: 10.1186/s12916-014-0238-7

Hughes MD, Bartlett RM (2002) The use of performance indicators in performance analysis. J Sports Sci 20(10):739–754. doi: 10.1080/026404102320675602

Hughes MD, Franks I (2005) Analysis of passing sequences, shots and goals in soccer. J Sports Sci 23(5):509–514. doi: 10.1080/02640410410001716779

Ingebrigtsen J, Dalen T, Hjelde GH, Drust B, Wisloff U (2015) Acceleration and sprint profiles of a professional elite football team in match play. Eur J Sport Sc 15(2):101–110. doi: 10.1080/17461391.2014.933879

IPTC (2001) SportsML. http://dev.iptc.org/SportsML

James N (2006) The role of notational analysis in soccer coaching. Int J Sports Sci Coach 1(2):185–198. doi: 10.1260/174795406777641294

James N, Mellalieu SD, Hollely C (2002) Analysis of strategies in soccer as a function of European and domestic competition. Int J Perform Anal Sport 2(1):85–103

Jones N (2014) Computer science: the learning machines. Nature 505(7482):146–148. doi: 10.1038/505146a

Jordan MI, Mitchell TM (2015) Machine learning: trends, perspectives, and prospects. Science 349(6245):255–260. doi: 10.1126/science.aaa8415

Kannekens R, Elferink-Gemser MT, Visscher C (2011) Positioning and deciding: key factors for talent development in soccer. Scand J Med Sci Sports 21(6):846–852. doi: 10.1111/j.1600-0838.2010.01104.x

Kasabian R (2014) World cup: assist goes to big data. Information week. http://www.informationweek.com/big-data/big-data-analytics/world-cup-assist-goes-to-big-data/a/d-id/1278822

Kempe M, Vogelbein M, Memmert D, Nopp S (2014) Possession vs. direct play: evaluating tactical behavior in elite soccer. Int J Sport Sci 4(6A):35–41

Kempe M, Grunz A, Memmert D (2015) Detecting tactical patterns in basketball: comparison of merge self-organising maps and dynamic controlled neural networks. Eur J Sport Sci 15(4):249–255. doi: 10.1080/17461391.2014.933882

Kihwan K, Grundmann M, Shamir A, Matthews I, Hodgins J, Essa I (2010) Motion fields to predict play evolution in dynamic sport scenes. Paper presented at the IEEE CVPR. 13–18 June 2010

Kim S (2004) Voronoi analysis of a soccer game. Nonlinear Anal Model Control 9(3):233–240

Knauf K, Memmert D, Brefeld U (2016) Spatio-temporal convolution kernels. Mach Learn 102(2):247–273. doi: 10.1007/s10994-015-5520-1

Kohonen T (1990) The self-organizing map. Proc IEEE 78(9):1464–1480

Kohonen T (2001) Self-organizing maps, 3rd edn. Springer, Berlin

Kong G, Xiao Z (2015) Protecting privacy in a clinical data warehouse. Health Inf J 21(2):93–106. doi: 10.1177/1460458213504204

Kostkova P, Brewer H, de Lusignan S, Fottrell E, Goldacre B, Hart G, Tooke J (2016) Who owns the data? Open data for healthcare. Front Public Health 4:7. doi: 10.3389/fpubh.2016.00007

Lago C (2009) The influence of match location, quality of opposition, and match status on possession strategies in professional association football. J Sport Sci 27(13):1463–1469. doi: 10.1080/02640410903131681

Lames M, McGarry T (2007) On the search for reliable performance indicators in game sports. Int J Perform Anal Sport 7(1):62–79

LeCun Y, Bengio Y, Hinton G (2015) Deep learning. Nature 521(7553):436–444. doi: 10.1038/nature14539

Lees A, Barton G (2003) A characterization of technique in the soccer kick using a Kohonen neural network analysis. J Sports Sci 22:491–492

Leser R, Baca A, Ogris G (2011) Local positioning systems in (game) sports. Sensors (Basel) 11(10):9778–9797. doi: 10.3390/s111009778

Levine TR, Hullett CR (2002) Eta squared, partial eta squared, and misreporting of effect size in communication research. Hum Commun Res 28(4):612–625. doi: 10.1111/j.1468-2958.2002.tb00828.x

Lewis M (2004) Money ball: the art of winning an unfair game. Norton & Company, Newyork

Liu H, Gomez MA, Lago-Penas C, Sampaio J (2015) Match statistics related to winning in the group stage of 2014 Brazil FIFA World Cup. J Sports Sci 33(12):1205–1213. doi: 10.1080/02640414.2015.1022578

Lohr S (2012) The age of big data. New York Times. http://www.nytimes.com/2012/02/12/sunday-review/big-datas-impact-in-the-world.html?_r=0

Lu WL, Ting JA, Little JJ, Murphy KP (2013) Learning to track and identify players from broadcast sports videos. IEEE Trans Pattern Anal Mach Intell 35(7):1704–1716. doi: 10.1109/TPAMI.2012.242

Lucey P, Bialkowski A, Carr P, Morgan S, Matthews I, Sheikh Y (2013) Representing and discovering adversarial team behaviors using player roles. Paper presented at the IEEE CVPR. 23–28 June 2013

Lucey P, Oliver D, Carr P, Roth J, Matthews I (2013) Assessing team strategy using spatiotemporal data. Paper presented at the 19th ACM SIGKDD, Chicago, Illinois, USA

Lynch C (2008) Big data: How do your data grow? Nature 455(7209):28–29

Mackenzie R, Cushion C (2013) Performance analysis in football: a critical review and implications for future research. J Sports Sci 31(6):639–676. doi: 10.1080/02640414.2012.746720

Marx V (2013) Biology: the big challenges of big data. Nature 498(7453):255–260. doi: 10.1038/498255a

McGarry T, Anderson DI, Wallace SA, Hughes M, Franks IM (2002) Sport competition as a dynamical self-organizing system. J Sports Sci 20:771–781

Medeiros J (2014) The winning formula: data analytics has become the latest tool keeping football teams one step ahead. Wired. http://www.wired.co.uk/magazine/archive/2014/01/features/the-winning-formula

Memmert D, Perl J (2009) Analysis and simulation of creativity learning by means of artificial neural networks. Hum Mov Sci 28(2):263–282. doi: 10.1016/j.humov.2008.07.006

Mesirov JP (2010) Computer science. Accessible reproducible research. Science 327(5964):415–416. doi: 10.1126/science.1179653

Mohr M, Krustrup P, Bangsbo J (2005) Fatigue in soccer: a brief review. J Sports Sci 23(6):593–599. doi: 10.1080/02640410400021286

Montoliu R, Martin-Felez R, Torres-Sospedra J, Martinez-Uso A (2015) Team activity recognition in Association Football using a Bag-of-Words-based method. Hum Mov Sci 41:165–178. doi: 10.1016/j.humov.2015.03.007

Moura FA, Martins LE, Anido Rde O, de Barros RM, Cunha SA (2012) Quantitative analysis of Brazilian football players’ organisation on the pitch. Sports Biomech 11(1):85–96. doi: 10.1080/14763141.2011.637123

Moura FA, Martins LE, Anido RO, Ruffino PR, Barros RM, Cunha SA (2013) A spectral analysis of team dynamics and tactics in Brazilian football. J Sports Sci 31(14):1568–1577. doi: 10.1080/02640414.2013.789920

Nakanishi R, Murakami K, Naruse T (2008) Dynamic positioning method based on dominant region diagram to realize successful cooperative play. In: Visser U, Ribeiro F, Ohashi T, Dellaert F (eds) Robo cup 2007: Robot Soccer World Cup XI, Vol 5001. Springer, Berlin, pp 488–495

Chapter   Google Scholar  

Nevill A, Atkinson G, Hughes MD (2008) Twenty-five years of sport performance research in the Journal of Sports Sciences. J Sport Sci 26(4):413–426. doi: 10.1080/02640410701714589

Noor AM, Holmberg L, Gillett C, Grigoriadis A (2015) Big data: the challenge for small research groups in the era of cancer genomics. Br J Cancer 113(10):1405–1412. doi: 10.1038/bjc.2015.341

Norton S (2014) Germany’s 12th man at the World Cup: Big Data. CIO Journal. http://blogs.wsj.com/cio/2014/07/10/germanys-12th-man-at-the-world-cup-big-data/

Ohmann C, Canham S, Danielyan E, Robertshaw S, Legre Y, Clivio L, Demotes J (2015) ‘Cloud computing’ and clinical trials: report from an ECRIN workshop. Trials 16:318. doi: 10.1186/s13063-015-0835-6

Olthof SB, Frencken WG, Lemmink KA (2015) The older, the wider: on-field tactical behavior of elite-standard youth soccer players in small-sided games. Hum Mov Sci 41:92–102. doi: 10.1016/j.humov.2015.02.004

Pääkkönen P, Pakkala D (2015) Reference architecture and classification of technologies, products and services for big data systems. Big Data Res 2(4):166–186

Passos P, Davids K, Araujo D, Paz N, Minguens J, Mendes J (2011) Networks as a novel tool for studying team ball sports as complex social systems. J Sci Med Sport 14(2):170–176. doi: 10.1016/j.jsams.2010.10.459

Perl J (2002) Game analysis and control by means of continuously learning networks. Int J Perform Anal Sport 2(1):21–35

Perl J (2004) A neural network approach to movement pattern analysis. Hum Mov Sci 23:605–620

Perl J, Weber K (2004) A neural network approach to pattern learning in sport. Int J Comput Sci Sport 3(1):67–70

Pikovsky A, Rosenblum M, Kurths J (2003) Synchronization: a universal concept in nonlinear sciences. Cambridge University Press, Cambridge

Pincus SM, Goldberger AL (1994) Physiological time-series analysis: what does regularity quantify. Am J Physiol 266:1643–1656

Rampinini E, Coutts AJ, Castagna C, Sassi R, Impellizzeri FM (2007) Variation in top level soccer match performance. Int J Sports Med 28(12):1018–1024. doi: 10.1055/s-2007-965158

Reed D, Hughes MD (2006) An exploration of team sport as a dynamical system. Int J Perform Anal Sport 6(2):114–125

Ric A, Hristovski R, Goncalves B, Torres L, Sampaio J, Torrents C (2016) Timescales for exploratory tactical behaviour in football small-sided games. J Sports Sci. doi: 10.1080/02640414.2015.1136068

Romanillos G, Zaltz Austwick M, Ettema D, De Kruijf J (2016) Big data and cycling. Trans Rev 36(1):114–133. doi: 10.1080/01441647.2015.1084067

Sampaio J, Macas V (2012) Measuring tactical behaviour in football. Int J Sports Med 33(5):395–401. doi: 10.1055/s-0031-1301320

Sampaio J, Lago C, Goncalves B, Macas VM, Leite N (2014) Effects of pacing, status and unbalance in time motion variables, heart rate and tactical behaviour when playing 5-a-side football small-sided games. J Sci Med Sport 17(2):229–233. doi: 10.1016/j.jsams.2013.04.005

Sarmento H, Marcelino R, Anguera MT, CampaniCo J, Matos N, LeitAo JC (2014) Match analysis in football: a systematic review. J Sports Sci 32(20):1831–1843. doi: 10.1080/02640414.2014.898852

Shafizadehkenari M, Lago-Penas C, Gridley A, Platt GK (2014) Temporal analysis of losing possession of the ball leading to conceding a goal : a study of the incidence of perturbation in soccer. Int J Sports Sci Coach 9(4):363–627

Shull PB, Jirattigalachote W, Hunt MA, Cutkosky MR, Delp SL (2014) Quantified self and human movement: a review on the clinical impact of wearable sensing and feedback for gait analysis and intervention. Gait Posture 40(1):11–19. doi: 10.1016/j.gaitpost.2014.03.189

Silva P, Travassos B, Vilar L, Aguiar P, Davids K, Araujo D, Garganta J (2014) Numerical relations and skill level constrain co-adaptive behaviors of agents in sports teams. PLoS One 9(9):e107112. doi: 10.1371/journal.pone.0107112

Sint R, Stroka S, Schaffert S, Ferstl R (2009) Combining unstructured, fully structured and semi-structured information in semantic wikis. Paper presented at the Semantic Wikis

Sitto K, Presser M (2015) Field guide to hadoop: an introduction to hadoop, its ecosystem, and aligned technologies. O’Reilly and Associates, Sebastopol

Taki T, Hasegawa J (2000) Visualization of dominant region in team games and its application to teamwork analysis. Proceedings of the paper presented at the computer graphics international, 2000

Tenga A, Holme I, Ronglan LT, Bahr R (2010a) Effect of playing tactics on achieving score-box possessions in a random series of team possessions from Norwegian professional soccer matches. J Sports Sci 28(3):245–255. doi: 10.1080/02640410903502766

Tenga A, Ronglan LT, Bahr R (2010b) Measuring the effectiveness of offensive match-play in professional soccer. Eur J Sport Sci 10(4):269–277. doi: 10.1080/17461390903515170

Toga AW, Dinov ID (2015) Sharing big biomedical data. J Big Data. doi: 10.1186/s40537-015-0016-1

Toga AW, Foster I, Kesselman C, Madduri R, Chard K, Deutsch EW, Hood L (2015) Big biomedical data as the key resource for discovery science. J Am Med Inform Assoc 22(6):1126–1131. doi: 10.1093/jamia/ocv077

Valter DS, Adam C, Barry M, Marco C (2006) Validation of Prozone ® : a new video-based performance analysis system. Int J Perform Anal Sport 6(1):108–119

Vogel H (1999) Gerthsen Physik, 20th edn. Springer, Berlin

Vogelbein M, Nopp S, Hokelmann A (2014) Defensive transition in soccer—are prompt possession regains a measure of success? A quantitative analysis of German Fussball-Bundesliga 2010/2011. J Sports Sci 32(11):1076–1083. doi: 10.1080/02640414.2013.879671

Waljee AK, Higgins PD (2010) Machine learning in medicine: a primer for physicians. Am J Gastroenterol 105(6):1224–1226. doi: 10.1038/ajg.2010.173

Wang Q, Zhu H, Hu W, Shen Z, Yao Y (2015) Discerning tactical patterns for professional soccer teams: an enhanced topic model with applications. Paper presented at the Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Sydney, NSW, Australia

Watts DJ, Strogatz SH (1998) Collective dynamics of ‘small-world’ networks. Nature 393(6684):440–442. doi: 10.1038/30918

Xinyu W, Long S, Lucey P, Morgan S, Sridharan S (2013, 26–28 Nov. 2013) Large-scale analysis of formations in Soccer. In: 2013 international conference on paper presented at the digital image computing: techniques and applications (DICTA)

Xue-wen C, Xiaotong L (2014) Big data deep learning: challenges and perspectives. Access IEEE 2:514–525. doi: 10.1109/ACCESS.2014.2325029

Yiannakos A, Armatas V (2006) Evaluation of the goal scoring patterns in European Championship in Portugal 2004. Int J Perform Anal Sport 6(1):178–188

Yu Y, Wang X (2015) World cup 2014 in the twitter world. Comput Hum Behav 48(C):392–400. doi: 10.1016/j.chb.2015.01.075

Yue Z, Broich H, Seifriz F, Mester J (2008) Mathematical analysis of a Soccer game. Part I: individual and collective behaviors. Stud Appl Math 121(3):223–243. doi: 10.1111/j.1467-9590.2008.00413.x

Zhang Y, Zhu Q, Liu H (2015) Next generation informatics for big data in precision medicine era. Bio Data Min 8:34. doi: 10.1186/s13040-015-0064-2

Download references

Authors' contributions

Both authors worked equally on all parts of the article. Both authors read and approved the final manuscript.

Acknowledgements

This work was supported by a grant from the German Research Foundation (Deutsche Forschungs Gesellschaft, ME 2678/3-3) to the second author.

Competing interests

The authors declare that they have no competing interests.

Author information

Authors and affiliations.

Institute of Cognition and Team/Racket Sport Research, German Sport University Cologne, Am Sportpark Müngersdorf 6, 50933, Cologne, Germany

Robert Rein & Daniel Memmert

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Robert Rein .

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License ( http://creativecommons.org/licenses/by/4.0/ ), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Reprints and permissions

About this article

Cite this article.

Rein, R., Memmert, D. Big data and tactical analysis in elite soccer: future challenges and opportunities for sports science. SpringerPlus 5 , 1410 (2016). https://doi.org/10.1186/s40064-016-3108-2

Download citation

Received : 11 May 2016

Accepted : 19 August 2016

Published : 24 August 2016

DOI : https://doi.org/10.1186/s40064-016-3108-2

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Sports performance
• Sports analytics
• Machine learning
• Spatiotemporal data
• Neural networks
• Deep learning
• Quantified self

• Frontiers in Sports and Active Living
• Elite Sports and Performance Enhancement
• Research Topics

Advances in the science of soccer: Challenges and Opportunities from the FIFA World Cup 2022

Total Downloads

Total Views and Downloads

About this Research Topic

The FIFA World Cup™ is probably the most popular sport event worldwide. The last edition, held in Qatar in 2022, presented several features that made it even more special. For example: • It was scheduled in November/December instead of the usual June/July due to the challenging ambient temperature ...

Keywords : soccer, football, FIFA, world cup, sport science, competition

Important Note : All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.

Topic Editors

Topic coordinators, recent articles, submission deadlines, participating journals.

Manuscripts can be submitted to this Research Topic via the following journals:

total views

• Demographics

No records found

total views article views downloads topic views

Top countries

Top referring sites, about frontiers research topics.

With their unique mixes of varied contributions from Original Research to Review Articles, Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author.

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Bioengineering (Basel)

Soccer Scoring Techniques—A Biomechanical Re-Conception of Time and Space for Innovations in Soccer Research and Coaching

Gongbing shan.

1 Biomechanics Lab, Faculty of Arts & Science, University of Lethbridge, Lethbridge, AB T1K 3M4, Canada

2 Department of Physical Education, Xinzhou Teachers’ University, Xinzhou 034000, China; nc.ude.utzx@gnaixgnahz

Xiang Zhang

Associated data.

Not applicable.

Background: Scientifically, both temporal and spatial variables must be examined when developing programs for training various soccer scoring techniques (SSTs). Unfortunately, previous studies on soccer goals have overwhelmingly focused on the development of goal-scoring opportunities or game analysis in elite soccer, leaving the consideration of player-centered temporal-spatial aspects of SSTs mostly neglected. Consequently, there is a scientific gap in the current scoring-opportunity identification and a dearth of scientific concepts for developing SST training in elite soccer. Objectives: This study aims to bridge the gap by introducing effective/proprioceptive shooting volume and a temporal aspect linked to this volume. Method: the SSTs found in FIFA Puskás Award (132 nominated goals between 2009 and 2021) were quantified by using biomechanical modeling and anthropometry. Results: This study found that players’ effective/proprioceptive shooting volume could be sevenfold that of normal practice in current coaching. Conclusion: The overlooked SSTs in research and training practice are commonly airborne and/or acrobatic, which are perceived as high-risk and low-reward. Relying on athletes’ talent to improvise on these complex skills can hardly be considered a viable learning/training strategy. Future research should focus on developing player-centered temporal-spatial SST training to help demystify the effectiveness of proprioceptive shooting volume and increase scoring opportunities in soccer.

1. Introduction

Soccer is the most popular sport in the World. Based on the information from Fédération Internationale de Football Association (FIFA), the game is played and watched on five continents with 265 million players and 4 billion fans, i.e., over 50% of the world population (7.7 billion) are linked to the game [ 1 , 2 , 3 ]. Yet, contrary to the popularity of the game, the number of scientific inquiries on key motor control skills, i.e., soccer scoring techniques (SSTs), appears disproportionately low when compared to the participation-to-scientific study ratios of other sports skills, such as complex gymnastics skills [ 4 ]. As a result, the scientific understanding of SSTs lags far behind its practice, with most participants acquiring various SSTs through individual experience rather than science-based instruction [ 5 , 6 ]. To make the matter worse, there is a dearth of scientific investigation on the many complex SSTs, such as the jumping turning kick, e.g., a nominated goal for the FIFA Puskás Award 2019, performed by Ibrahimović [ 7 ] and the diving scorpion kick, which won the FIFA Puskás Award 2017 [ 8 ]. These SSTs seem virtuosic in scope and are commonly believed to be the talent ability solely of soccer stars [ 9 ]. Obviously, relying on the aptness of the athletes to improve these virtuosic SSTs can hardly be considered a viable learning or coaching strategy. Worst of all, the terms surrounding SSTs used in practice and research have been confusing [ 10 ]; researchers and practitioners are not sure how many SSTs are available for the game. This scenario hinders not only the scientific studies on many exceptional SSTs but also the development of novel coaching methods for learning these SSTs.

It is well known that the greatest attraction of soccer is goal scoring. Compared to many other sports, goals are relatively rare in soccer–on average, less than three goals per game in FIFA world cups since the 1960s [ 11 ]. Because of their rarity, soccer goals are extremely exciting for millions of fans. The various means by which gameplay moves toward a goal can be thought of as an improvised drama, where emotional tension is built over long periods only to be fully released when the goal is achieved. This characteristic contributes to making soccer the most popular spectator sport in the world. Therefore, an essential core of soccer research and coaching is to help athletes master aesthetically eye-catching SSTs for increasing both scoring possibilities and the excitement of the game.

1.1. Novel Scientific Studies Required for the World’s Most-Popular Sport

Through systematic identification, a recent study [ 10 ] has revealed that there are 43 SSTs that exist in current soccer games. Surprisingly, there are only a handful of SSTs documented in the existing training/coaching literature and, consequently, the comprehensive list of SST training is presently limited to maximal instep kick (including the curled kick), jumping headers, and volley/side volleys [ 12 , 13 , 14 ]. Since the existing scientific understanding could not supply enough help to practitioners in developing training methods for learning more SSTs, practical modi operandi have been the main driving force for keeping SST exuberant.

At the practical level, celebrations of the goal of the month, of the season, of the year, and of the decade, are widely adopted in professional soccer leagues for recognizing players who are deemed to have scored the “most beautiful goal” [ 10 ]. Those glorious goals are normally chosen by a combination of panel experts and a public vote. This practice has been successfully used to encourage and promote aesthetically significant goals from players, as it sparks the creativity of the athletes to develop novel/unique SSTs through “self-learning”. Even infrequently, under this practical modi operandi , exceptional SSTs have been developed by a few talented athletes from time to time. These virtuosic skills have become models for more athletes and coaches to impersonate and duplicate blindly, i.e., learning without insight knowledge obtained from research. Especially the brilliant goals scored in soccer’s flagship tournaments, such as those held by FIFA, UEFA (Union of European Football Associations) and other professional national soccer leagues with an international reputation (e.g., La Liga/Spain, Serie A/Italy, Bundesliga/Germany, Premier League/England, and Ligue 1/French), are mimicked by worldwide soccer players. Biomechanically, there are two issues related to such blind self-learning: learning efficiency and injury risk.

Soccer shots, like many complicated human movements, are trained motor skills. Studies have demonstrated that systematic and scientific inquiry into the biomechanics of human motor skills has great potential to demystify complicated human motor skills for efficient learning [ 15 , 16 , 17 , 18 ]. Hence, knowledge obtained from biomechanics studies can help optimize performance outcomes while simultaneously reducing the risk of training-related injury. There are two useful biomechanical analyses: kinematic and kinetic quantification of motor skills. In terms of motor learning, kinematics has significant utility in terms of improving teaching and learning, while an understanding of kinetics is essential for reducing the risk of learning and playing-related injury. Both kinematics and kinetics have significant utility in raising practitioners’ awareness of biomechanical “cause and effect”, thus giving them additional tools and knowledge to optimize their practice [ 6 , 17 , 19 , 20 , 21 ].

From a motor learning point of view, one common criterion of the honored goals in soccer’s flagship tournaments is their repeatability, e.g., the FIFA Puskás Award states that “the goal should not be the result of luck or mistakes by the other team” [ 22 ], suggesting that the nominated goals and the related SSTs are theoretically repeatable, i.e., entrainable. What is needed for establishing a scientific training system is the knowledge of the biomechanical parameters that influence the quality of the soccer shots. Unfortunately, most of the 43 SSTs [ 10 ] identified so far are “off the radar” to researchers. The SSTs overlooked in research are normally airborne and/or acrobatic, perceived as high-risk and low-reward. Counting on athletes’ talents to improvise on those SSTs is neither scientific nor realistic. Hence, novel biomechanical studies are needed for the scientific discovery of these airborne and/or acrobatic SSTs. As a result, the skills that are virtuosic in appearance may be eminently trainable and less a product of the improvisatory abilities of individual players.

1.2. Literature Review

Web of Science is a reputable resource due to its guaranteed proofed scientific content [ 23 ]. Among all types of papers, review articles draw upon published articles to provide a great overview of the existing literature on a topic. As such, to reveal the current state of scientific studies on soccer scoring with rigorous and quality information, the Web of Science database was searched for systematic review articles in December 2021. The search was performed by using the keywords “football” and “soccer“, with each associated with the terms “goal analysis” and “review”. The initial search identified 31 papers in the database. The titles and abstracts of 31 articles were then screened according to their relevance to SSTs in elite soccer, resulting in 27 studies being eliminated. At the end of the screening procedure, four systematic review articles [ 24 , 25 , 26 , 27 ] between February 2018 and January 2020 received further in-depth reading to identify the current research state in this area. The articles show that soccer scoring quantification started as early as 1968, investigating the statistical relationship between the number of passes and goals scored by analyzing 3213 professional matches from 1953 to 1968 [ 28 ]. Over the past half-century, more and more notational measurements were developed and applied to finding the factors influencing soccer goal scoring. These measurements include total passing frequencies [ 29 ], ball possession time [ 30 , 31 ], shots at goal [ 32 , 33 ], the position of an attempt on goal (e.g., three longitudinal areas: right, center, and left, and zone: ultra-defensive, defensive, central, offensive, and ultra-offensive) [ 34 , 35 ], the influence of set plays or dead ball routines [ 36 , 37 ] etc. In short, the previous studies on soccer goals have overwhelmingly focused on the development of goal-scoring opportunities or game performance analysis, i.e., tactical strategies, and overlooked the role of various SSTs on soccer goals.

As matter of a fact, very few studies have analyzed the special kicking skills that elite goal scorers possess [ 9 , 27 ]. As revealed by the recent systematic review article, the existing research studies have not fully explored the final actions of the players in goal situations [ 27 ]. There is no doubt that the current studies can provide important references for coaches in designing training programs for developing the tactical strategies of a team. However, they could not improve the SST training. It is well known that all tactical strategies learned and trained aim at achieving the ultimate shot opportunity, and the key to the success or failure of the shot depends largely on the SST selected by the player.

A study on defense in all 306 German Bundesliga games from the 2010/2011 season [ 38 ] showed that the top teams have a faster defensive reaction time compared to the remaining teams. The result suggests that the time taken to perform a shot is becoming shorter and shorter as the competition level increases. FIFA has vividly described this development trend as “every nanosecond is special” [ 39 ]. As a consequence, more and more airborne shots are seen in current elite games, and the height of an airborne kick is increasing higher and higher in the top elite-level games. Unfortunately, both current research and existing training systems can no longer keep up with actual real-game development.

1.3. The Current State of SST Knowledge

In this vein, few existing studies have found that shots with the feet seem to achieve between 70–80% of the goals, whereas the rest of the goals are achieved from headers [ 40 , 41 ]. More details have been added by a recent study [ 10 ], finding that there are 43 SSTs identified in elite soccer games, and over 60% of them are airborne SSTs. The airborne shots contribute to over 50% of goals, with shots attempted by foot, head, or chest. These findings indicate that scoring opportunity identification has to consider factors linked to airborne shots. Sadly, the most up-to-date systematic review article on the biomechanics of kicking in soccer is a dozen-years old, which focuses only on the instep kick [ 42 ], with a dearth of scientific studies on most of the airborne SSTs [ 4 , 43 ].

There are also limited studies on the anatomic parts of the foot used to shoot. One study has found that the instep seems to be the most used anatomic part, followed by the inside part of the foot [ 44 ]. The recent study conducted by Zhang and Shan [ 10 ] added more details to this; depending on the scorer’s body posture when shooting (e.g., facing, side-facing, or back-facing) and ball position (e.g., ground or airborne), the instep, dorsi-side, inside, outside, toe, heel, or plantar-side are selected by scorers for shots.

Due to the limited knowledge related to SSTs, there is a scientific gap in current scoring-opportunity identification. From a scientific standpoint, both temporal and spatial variables must be examined when evaluating scoring opportunities. In essence, scoring chances are an issue of maximizing the probabilities of scoring, which cannot be determined without basic theories and knowledge related to the temporal-spatial identification and quantification of a player’s possibilities to shoot.

From the temporal perspective, it is well known that even if in possession of a free ball, a player will likely not be free for long; defenders will attempt to thwart the shot. A representative study [ 40 ], which analyzed all goals in the English Premier League during the 2008/2009 season, found that scoring with zero possession (i.e., a “one-touch shot” where a player shoots as a ball is passing by) accounts for 69.3% of goals. Setting up a shot with one or more contacts of the ball results in only 17.9% of goals. These results indicate that the longer a player possesses the ball, the lower the scoring chance. The least favorable situation for scoring is when players score after individual dribbling (12.8%).

Regarding spatial variables, existing studies have shown that 65–90% of goals are scored when an attacking player possesses the ball in or near the opponents’ penalty area, and he/she is not hindered by defenders, i.e., a “free ball” [ 44 , 45 , 46 , 47 ]. The studies have actually confirmed the practitioners’ empirical evidence. This explains why most of the current goal-scoring training overwhelmingly emphasizes the geographic location of the ball on the playing field.

In brief, the present research scheme fails to consider the player-centered temporal-spatial aspects of SSTs and underemphasizes air-attack (3D consideration) related to shooting techniques.

1.4. Research Aims and Research Questions

The current scenario would suggest that ground-breaking research is needed in order to develop science-based SST training regimes for improving scoring possibility. The current paper aims to lay a foundation for launching a ground-breaking study via the re-conception of temporal and spatial factors related to soccer shooting, identifying elements that could be applied in the entrainment of complex SSTs via biomechanical quantification. The study has three specific research questions to be answered (i.e., the main goals of the current study):

• Scientifically, what is the content of temporal–spatial opportunity identification related to SST?
• What is a quantitative and reliable method to evaluate the improvement of scoring possibility through science-based training?
• What is the theoretical implication of these findings related to the innovative development of SST training?

2. Materials and Methods

For answering these research questions, the current study initiates a novel theoretical framework, which has its origin in elite soccer. Three steps are involved in the establishment of the original framework: (1) selection of temporal criteria for efficiency recognition, (2) identification of new/potential spatial variables via the video-based analysis of all 132 nominated goals of the FIFA Puskás Award between 2009 and 2021 [ 22 ], and (3) quantification of scoring possibility via biomechanical modeling.

2.1. A New Theoretical Framework for Developing SST Training—Focusing on Time in Space

Scoring opportunity can be defined by the feasibility of shooting successfully. Practically, it is well known that when an attacker gets a scoring opportunity, his/her chance to shoot the ball typically does not last very long. In these brief moments, a player needs to shoot the ball quickly (the temporal aspect of goal scoring) and accurately (the spatial feature of goal scoring). The reality in elite games is actually more complicated than this simplified thinking. Regardless of the geographic opportunities one could gain, only considering the dynamic relationship between an attacker, the goal, and the 3D position of the ball related to the scoring chance identification, is not enough to reach a decision without knowing the biomechanical characteristics of various SSTs. Therefore, a new framework should link the temporal and spatial details to the SSTs for ameliorating the scoring possibility.

Since there is no existing framework for considering the temporal and spatial aspects of SSTs simultaneously, our study will bridge the gap by creating a new theoretical framework. The original framework covers factors related to the temporal efficiency and spatial effectiveness of SSTs. The basic ideas of the framework are: (1) to mathematically evaluate the effective shooting volume for scoring chance quantification, i.e., a quantitative determination of spatial feasibility for shooting, and (2) for any dynamic ball (chance) covered by the effective shooting volume, to choose a proper SST for shooting without delay, i.e., temporal efficiency of shooting. The novel aspect of the original framework is to introduce 3D considerations (of player-centered temporal-spatial consideration) for the training of SSTs in order to improve scoring possibility.

The key element of the framework is the effective shooting volume. Its quantification will inevitably require the study of the overlooked airborne and/or acrobatic SSTs. These SSTs are intricate, belonging to gymnastic-like motor skills. Previous studies have revealed that the learning quality of gymnastic-like motor skills depends on one’s proprioceptive ability and can be progressively developed through structured repetitive training [ 48 , 49 , 50 , 51 ]. Hence the effective shooting volume can also be named as the player’s proprioceptive shooting volume. This volume determination would fundamentally inform if a dynamic ball should be counted as a goal chance (i.e., the ball falls within or passes through the volume) or not.

In summary, the core of the novel framework is to create a new quantitative way for (1) enlarging the effective shooting volume to cover more goal chances and (2), at the same time, enhancing the shooting ability to ensure one-touch shots within this volume. Based on these core elements, this original framework can be denominated as “Focusing On Time In Space”. It aims to nexus the temporal efficiency and spatial effectiveness of maximizing soccer scoring possibilities. Whence, it would build a scientific foundation for innovations in future SST research and training.

2.2. Selection of Temporal Criteria

Our selection of temporal criteria is based on a study related to the temporal efficiency of SSTs carried out by Durlik and Bieniek (2014) [ 40 ]. Since players scoring after individual dribbling involves other motor skills rather than SST skills only, a modification has been made, i.e., the scoring after individual dribbling was excluded. Additionally, any further actions before shooting have been proven to decrease scoring possibility [ 40 ]; therefore, it is logical to select the amount of possession before shooting so as to recognize overall temporal efficiency. Hence, the following categories were selected for the quantification in this study:

• one-touch-shot (zero-possession), where a player shoots as a ball is passing by;
• One possession, i.e., setting the ball and then shooting;
• the other ball control strategies combined, i.e., 2+ possession maneuvers.

The temporal criteria were evaluated via the statistical results of the 132 Puskás nominated goals between 2009 and 2021 [ 22 ].

2.3. Identification of New Spatial Variables

Regarding spatial analysis, previous studies have focused only on field geography [ 44 , 45 , 46 , 47 ] and have neglected those factors related to SSTs, e.g., the athlete’s body orientation facing, side-facing, or back-facing the goal, and the spatial position of the ball at the instance of a shot [ 10 ]. To give more detail to the latter aspect, the spatial position of the ball has horizontal and vertical components. Horizontally, using the goal and the player as positional references, the ball can be between them, beyond them, or to the side of them. Vertically, the ball can be airborne or on the ground ( Figure 1 ). Therefore, in the current study, these new spatial parameters related to SSTs at the instance of a shot were induced and designated for quantitative analysis.

The identification and clarification of spatial factors. ( A ) The 3D quantification of a jumping side volley–top view [ 4 , 43 ]. ( B ) The 3D quantification of the maximal instep kick–top view [ 52 , 53 , 54 ]. ( C ) The 3D quantification of the bicycle kick–top view [ 9 , 43 ].

Combinatorically, the above selected temporal and spatial parameters were applied to the quantitative examination of all 132 Puskás shooting videos. Descriptive statistics (i.e., pie chart) were applied to summarize these categorical data (i.e., percentage distribution) in order to reveal the characteristics of the selected parameters and their contributions to scoring goals at the top elite level.

2.4. Quantification of the Proprioceptive Shooting Volume via Biomechanical Modeling

Dimensional data obtained from 3D motion capture and/or full-body biomechanical modeling were used to estimate the proprioceptive/effective shooting volume. The quantification works as follows:

• Apply 3D biomechanical modeling to quantify various kicking techniques [ 4 , 9 , 53 , 55 ];
• Target and select three SSTs to determine the anterior–posterior, medial–lateral, and vertical dimensions when the kicks are being performed. The selection of SSTs represents the shooting ability that an athlete can obtain after the current training methods or that talented elite players have already performed in elite soccer;
• Use the three dimensions of the selected SSTs to estimate the proprioceptive shooting volumes of the two different performance levels and to determine the difference between them.

The 3D dimensional data used in this study are either from the authors’ previous 3D quantification studies or gained through both anthropometrical study [ 56 ] and biomechanical model estimation/simulation. Clearly, different selections of SSTs will result in distinct sizes of the volume, i.e., various SSTs will lead to a differentiation in scoring possibility. Logically, the larger one’s effective shooting volume, the more goal chances one will possess. Through the novel framework, the current article has, for the first time, visualized the concept of how to establish a link between the temporal-spatial aspects of SSTs and goal chance quantification. This conceptual exploration is currently missing in soccer research and practice.

The analysis results of temporal and spatial factors related to goal-chance identification are shown in Figure 2 . Compared to the outcomes of temporal efficiency from the previous study [ 40 ], i.e., 69.3% of goals scored for zero-possession shots, 17.9% of goals scored after two or more possessions, and 12.8% of goals scored for shots after dribbling, the temporal branch of Figure 1 based on the 132 FIFA Puskás Award nominated goals shows comparable results: 56.8% (0 possession), 70.5% (0 & 1 possession), and 29.5% (2 + possessions), respectively.

The temporal and spatial factors related to goal chance identification and their contributions to goals: evidence drawn from all 132 FIFA Puskás Award nominees’ goals, 2009–2021.

The results of the spatial analysis of the 132 Puskás goals denoted that, as a ball comes into an athlete’s proprioceptive shooting volume, a player’s body orientation can be facing, side-facing, or back-facing ( Figure 1 and Figure 2 ) relative to the position of the goal. In terms of body orientation, at the moment of shooting among the 132 Puskás goals, about half (51.5%) were achieved facing the goal, while 31.8% were “side-facing”, and 16.7% were “back-facing” ( Figure 2 ). The statistics for horizontal ball position are comparable, with 53.0% of goals having occurred when the ball was between the player and goal, while side and beyond balls accounted for the remaining 47.0%. Regarding vertical ball positioning, 56.1% of the goals were airborne, and only 43.9% were ground balls.

The three-dimension comparisons between the SSTs trained by the current system and the selected SSTs performed by elite players are shown in Figure 3 . As elaborated in the method section, these three-dimensional values were applied to quantitatively estimate the shooting volume formed by skills in terms of a player’s body height (BH). It should be noted that (1) the estimation represents the maximum volume one could reach, and (2) the dimensions normalized by BH increased the generalization of the method [ 55 ]. In current soccer coaching practice, the systematically trained SSTs are the maximal instep kick (including the curled kick), jumping headers, and volleys and side volleys [ 12 , 13 , 14 ]. Therefore, the proprioceptive shooting volume of the currently-trained SSTs could be calculated by the following dimensions:

• Anterior–posterior dimension is 1.3 BH, i.e., the last-stride length of the maximal instep kick [ 52 , 54 , 55 ] ( Figure 3 A);
• Medial–lateral dimension is 0.8 BH, i.e., the lateral reach of the side volley of 0.4 BH by each leg ( Figure 3 B);
• Vertical dimension of 1.4 BH, i.e., the jumping height for a header ( Figure 3 C).

The dimensions of selected SSTs and their influences on the quantification of the attack space (scoring chance): ( A ) The 3D quantification of the maximal instep kick–side view with timely trace of the kicking foot, obtained via 3D-motion analysis and biomechanical modeling [ 52 , 53 , 54 ]. ( B ) Lateral dimension of volley kick–distance estimated using biomechanical modeling and anthropometrical study [ 5 , 56 ]. ( C ) Vertical dimension of a jumping header–distance estimated using jump biomechanics [ 57 , 58 ] and anthropometrical modeling [ 56 ]. ( D ) Frontal dimension of a diving header–distance estimated using diving biomechanics [ 59 ] and anthropometrical modeling [ 56 ]. ( E ) The 3D quantification of the bicycle kick–side view with timely trace of the kicking foot, obtained via 3D-motion analysis and biomechanical modeling [ 9 , 43 , 56 ]. ( F ) The 3D quantification of jumping side volley–frontal view with timely trace of the kicking foot, obtained via 3D-motion analysis and biomechanical modeling [ 4 , 43 ]. ( G ) Potential attack height reached by using a bicycle kick at different trunk-angle orientations–biomechanical model estimation [ 9 , 56 ].

In the case of a 1.8 m tall athlete, when the above dimensions are considered, the calculation of his/her maximum effective shooting volume would be 2.34 m × 2.52 m × 1.44 m, or very roughly 8.5 m 3 .

A similar quantitative estimation was also applied to the selected SSTs performed by elite players. The practical potential of the quantification of the proprioceptive/effective shooting volume is shown in Table 1 . For the same case (1.8 m tall player), if the volumes of two acrobatic SSTs (i.e., bicycle kick and jumping side volley, Figure 3 E,F, performed by talented elite soccer players) are included in the calculation, the same player’s effective shooting volume would be 4.14 m × 2.52 m × 3.24 m, roughly 35.4 m 3 , or approximately four times the shooting volume of the normally practiced techniques in current training practice. If additional SSTs, such as the diving header ( Figure 3 D) and an improved bicycle kick (trunk angled 45°, in addition to the current 0° bicycle kick, Figure 3 G) are included, the effective shooting volume increases to 4.32 m × 2.70 m × 5.04 m, roughly 58.8 m 3 , or approximately seven times the normally practiced shooting volume. Ad hoc, more quantitative estimations could be performed, e.g., the long-jump header [ 10 ] performed by Cristiano Ronaldo in the 2008/2009 UCL season ( Figure 4 ). If this SST could be quantified, the effective shooting volume would definitely increase further.

The long-jump header (the player is identified in red circle on the two left frames). Currently, there is no study available for revealing the 3D dimension of this SST (the figure was generated from the video of UEFA TV video, published in 2021 [ 60 ]).

The comparison of the proprioceptive/effective shooting volume from the current practice (100%) to the player-centered temporal-spatial training of airborne/acrobatic SSTs.

4. Discussion

Due to the rarity of goal chances in soccer, it is highly relevant to establish a theoretical system for innovating research and training practice around goal scoring. Obviously, the re-conception of scoring chance quantification and its relationship to the temporal and spatial aspects of SSTs would have great potential to form a foundation on which novel developments of various SSTs could be built. Unfortunately, scientific studies on the influence of temporal-spatial factors on shooting effectively (i.e., quantify the scoring chance and turning that scoring chance into a goal) have not yet been conducted. This is why our study proposed a novel framework for this area. Based on the results, this section makes an effort to answer research questions 1–3 to elaborate the conceptual metaphors of the novel framework.

4.1. Temporal-Spatial Opportunity Identification

The evidence gained from the 132 FIFA Puskás goals has clearly indicated that a new theoretical framework is needed for re-examining and/or allocating the known and unknown factors related to existing SST, and breakthrough studies are desirable to quantify the influences of those factors on temporal-spatial opportunity identification.

4.1.1. Temporal Efficiency

The first challenge would be a temporal enumeration, i.e., the number of possessions of the ball during preparation for shooting. As portrayed before, even if in possession of a “free ball”, a player will likely not be free for long; defenders will attempt to thwart the shot. Therefore, any delay in preparing a shot will decrease temporal efficiency. The results summarized in this study clearly indicate that a sudden attack would create the highest temporal efficiency for converting a scoring chance into a goal. This result is comparable to the representative study [ 40 ] on the analysis of all goals from the English Premier League during the 2008/2009 season. Yet, the temporal aspect does not stand alone, and obviously, it interacts with spatial factors.

4.1.2. Spatial Effectiveness

The special aspect presents a more challenging concept, but one thing is certain: it must consist of more than mere field geography. A player’s proprioceptive abilities influence goal-opportunity identification, i.e., as the soccer ball comes into a player’s proprioceptive shooting volume, his/her ability to perform a shot under various body orientations with the 3D spatial positions of the ball must fall within these abilities in order for him/her to attempt to score. Current knowledge shows that proprioceptive abilities can be developed through repetitive training through structured and targeted training [ 48 , 49 , 50 , 51 ].

Yet, spatial factors may be over-simplified in the present training practice. There seems to be an existing research and coaching emphasis on shots taken (1) facing the goal and (2) with the ball between the player and the goal [ 12 , 13 , 14 ]. It is not surprising that most training and practice regimes concentrate on the variables that account for the higher percentages of goals found in the pie charts of Figure 2 . However, this neglects a significant percentage of the scoring opportunities that may exist in smaller “slices of the pie”. Some of these might be improved through scientifically structured training. For example, Swedish player Ibrahimovic’s bicycle kick against England in 2013 (i.e., the Puskás award 2013 [ 61 ]) was a goal achieved as he ran away from, and not toward, the goal. This shot can be characterized as follows: 0 possession, back facing the goal, a beyond ball, and an air attack. Currently, there is no science-based training program available to practitioners for learning this skill [ 9 , 43 ]. Therefore, the skill is generally considered a product of the improvisatory and virtuosic abilities of an individual player, not a trainable one. The question arises whether or not the training regimes that develop the SST shooting components required for such “virtuosity” might improve scoring percentage by increasing a player’s proprioceptive/effective shooting volume. The 132 Puskás goals have shown that 16.7% of the goals were scored by using back-facing goal techniques, and 19.7% of the goals were scored with the ball located beyond (not between) the goal and the players ( Figure 2 ). These “unusual” goals would signify that what is required is a new theoretical framework to logically link the temporal and spatial factors for a systematical exploration of these acrobatic skills for developing science-based training methods.

4.1.3. Lost in Time and Space in the Current Scoring Research

The current study reveals that joint consideration of the temporal and spatial aspects of scoring has been mostly neglected in the current SST research and coaching literature, i.e., scientific studies on SST could be considered lost in time and space. The results of Puskás goals imply that a few talented elite athletes, via years of practice, have magically linked temporal efficiency and spatial effectiveness in developing improvisatory abilities. Such abilities can turn “impossible” goals (commonly identified as non-chance goals) into goals. The scientific re-conception considered in this study would suggest that the biomechanical quantification of the temporal–spatial factors of these extraordinary SSTs could help us demystify these “magic kicks”, and evidently identify a scoring chance as “yes” or “no” instead of “impossible”. Without the scientific quantification of these temporal–spatial challenges, the identification of a scoring chances is vague, subjective, and unclear. Unfortunately, this is the situation for the current studies on soccer scoring, especially for the spatial challenges related to SSTs.

In summary, the current study is the one that has attempted to originate a novel framework for scientifically pinpointing the content of temporal–spatial opportunity identification. As such, future innovations in SST research and training system development could be launched by applying the proposed framework.

4.2. Quantification of Athletes’ Proprioceptive/Effective Shooting Volume—A Key for Scoring-Opportunity Identification

Scientifically, the quantification of a player’s attack volume would impartially show whether or not a ball should be counted as a chance of a goal. At the present time, there are few, if any, studies on the training manipulation and expansion of a player’s effective shooting volume. Within the current knowledge, the training effect on the volume depends on the learner’s spatial motor control ability, i.e., the proprioceptive competence, which is highly entrainable [ 48 , 49 , 50 , 51 ]. Up to now, this volume is still limited by the following SSTs in coaching practice [ 12 , 13 , 14 ]:

• Maximal instep kick (including curled kicks), characterized as kicks facing the goal, between balls, and ground balls;
• Headers, characterized as shots facing/side-facing the goal, between/side balls, and an air attack;
• Volleys, characterized as kicks facing/side-facing the goal, between/side balls, and an air attack below the hip.

Therefore, the current training system would under develop a player’s proprioceptive/effective shooting volume, reducing their possibilities to shoot and decreasing their goal-scoring chances.

The results of this study clearly demonstrated that the limited proprioceptive shooting volume is a result of the insufficient SSTs with which a player has been entrained. The results of Table 1 would suggest that talented athletes, e.g., Cristiano Ronaldo and Zlatan Ibrahimović (both winners of the Puskás Award and are able to perform a bicycle kick and a jumping side volley), have found ways to master unique acrobatic SSTs for increasing their proprioceptive/effective shooting volume by four times that of normally trained players. Evidently, these unusual SSTs contribute to enlarging their proprioceptive shooting volume. Furthermore, this study has revealed that, theoretically, the increased volume could potentially still be enlarged to seven or more times that of the “normally” practiced shooting volume ( Table 1 ). The ramifications of our results are far-reaching. Even if less than half of these theoretical gains could be realized through the training of the underutilized “slices of the pie” from Figure 1 , a player’s effective proprioceptive shooting volume could be doubled or tripled.

In short, to maximize scoring probabilities, it is essential to intensely expand the dimensions of the proprioceptive shooting volume, which is heavily influenced by the SSTs available to an athlete. The goals of Puskás nominees unveil that there are more than a dozen SSTs that should be considered as potential skills to maximize a player’s proprioceptive shooting volume. Based on FIFA criteria [ 22 ], these SSTs are undeniably repeatable; as such, they should be entrainable. What we need are scientific studies that demystify these SSTs and establish new and effective training regimes for developing these extraordinary SSTs among young and future players.

4.3. Focusing on Time in Space—The Nexus for Uniting Time Efficiency and Spatial Effectiveness

The nexus for uniting the time efficiency and spatial effectiveness of players should be rooted in the development strategy of future SST training. The logical sequence would be:

• Firstly, master as many SSTs as possible through science-based SST training. This step would aim at enlarging the effective 3D dimensions of the proprioceptive shooting volume, i.e., increasing spatial effectiveness. This is the foundation for maximizing scoring possibilities;
• Secondly, entraining for decision ability in properly selecting an SST for an accurate attack by means of shooting various dynamic balls. This step would focus on temporal efficiency. Mastering more SSTs would allow a player to choose the proper SST for ensuring a zero-possession shot, regardless of the ball’s horizontal and vertical position as well as an attacker’s dynamic posture ( Figure 1 , Figure 2 and Figure 3 );

The scorpion kick, a nominated goal of FIFA Puskás Award 2014 performed by Ibrahimović (the figure was generated from the video of FIFA Puskás Award 2014).

Regrettably, scientific studies, as well as coaching practice, fall far behind in initiating the above training regime. Most of the identified SSTs (currently 43 in total [ 10 ]) cannot be systematically trained due to the lack of scientific investigation and understanding. As elaborated before, the SSTs overlooked by researchers and the current training regime are the airborne and/or acrobatic attack techniques, such as the scorpion kick, diving scorpion kick, jumping side volley, jumping turning kick, long-jump turning header, diving header, sliding kick, rabona kick, and more [ 10 ]. The fatal effect of the overlook is the decrease in time efficiency and spatial effectiveness.

One common characteristic of the SSTs overlooked in research and coaching practice is exceedingly complex, labeled as high-risk and low-reward [ 9 ]. These skills, for most athletes, cannot be learned without insightful guidance. It is time for researchers to conduct quantitative studies for (1) demystifying the complexity of the skills, (2) identifying the skills required for various ball positions and dynamic body postures, and (3) develop training programs for injury-free exercise, since the accurate performance of the airborne and/or acrobatic SSTs requires repetitive training.

This is the first study on the time and space aspects of SSTs. It is understandable that there are limitations associated with this study. There are two obvious ones. First, due to the unavailability of the 3D data of most SSTs, the current quantification on the spatial effectiveness can only be used as a reference for practitioners. More future studies are inevitably needed for reaching an accurate result. Second, there might be gender-based control-pattern variations. A quantitation of female athletes, as well as comparisons between the males and the females, must be conducted in the future studies.

Summarized above, the practical information for coaches and players is as follows: (1) the more SSTs an athlete can perform, the more goal chances he will have; and (2) the more airborne SSTs an athlete can master, the more efficient and effective his shooting attacks will be. Yet, merely relying on the aptness of an athlete to master these extraordinary SSTs would be hit or miss. Such an approach cannot be considered as a viable coaching strategy. A more science-structured learning, based on the re-conceptual organization of temporal-spatial aspects of SSTs should be developed. The development ought to firstly focus on improving players’ spatial awareness in terms of body orientation and ball spatial position and, subsequently, increase the temporal efficiency through practices of a one-touch-shot within this improved volume. In short, focusing on Time-in-Space would be the nexus for uniting time efficiency and spatial effectiveness in future SSTs learning and training.

5. Conclusions

Retrospectively, from a scientific standpoint, both temporal and spatial variables must be examined when evaluating soccer scoring opportunities. Unfortunately, field geography (i.e., the development of goal scoring opportunities) or game analysis in elite soccer tends to dominate the attention of researchers and practitioners and consideration of the player-centered temporal-spatial aspects of SSTs is mostly neglected.

Goal scoring research fails to address the time and space related to SSTs. Space certainly consists of more than mere field geography. A player’s trained SSTs influence both scoring opportunity identification and the dimensions of his/her attack space. The development of novel training programs should, first, focus on the increase of the proprioceptive shooting volume through mastering as many SSTs as possible; then, it should concentrate on the training of selecting a proper SST to reach one-touch-shots within the enlarged attack volume, i.e., focusing on time in space.

The current study reveals that a player has to learn airborne and acrobatic SSTs in order to increase his/her spatial effectiveness, as well as the temporal efficiency of shooting. Therefore, scientific studies are indisputably needed to demystify the complexity of these skills for developing their learning and training.

The great attraction of soccer for millions of fans is the goal. Various techniques for scoring goals are sources of excitement. More frequent use of airborne and acrobatic SSTs for goals can only enhance the excitement of the game. Therefore, this new theoretical framework would bring more excitement by promoting novel studies and developing innovative training programs for learning and practicing various SSTs.

Funding Statement

This research was funded by Discovery Development Grant of National Sciences and Engineering Research Council of Canada (NSERC), Grant #: DDG-2021-00021.

Author Contributions

Both G.S. and X.Z. are actively involved in the conceptualization, methodology, validation, formal analysis, investigation, resources, writing—original draft preparation, review and editing, as well as visualization of the paper. G.S. is for the funding acquisition. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

Informed consent statement, data availability statement, conflicts of interest.

The authors declare no conflict of interest.

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Soccer Athlete Data Visualization and Analysis with an Interactive Dashboard

• Conference paper
• First Online: 29 March 2023
• Cite this conference paper

• Matthias Boeker   ORCID: orcid.org/0000-0001-6217-2658 15 , 16 &
• Cise Midoglu   ORCID: orcid.org/0000-0003-0991-4418 15 , 17  

Part of the book series: Lecture Notes in Computer Science ((LNCS,volume 13833))

Included in the following conference series:

• International Conference on Multimedia Modeling

1487 Accesses

Soccer is among the most popular and followed sports in the world. As its popularity increases, it becomes highly professionalized. Even though research on soccer makes up for a big part in classic sports science, there is a greater potential for applied research in digitalization and data science. In this work we present SoccerDashboard , a user-friendly, interactive, modularly designed and extendable dashboard for the analysis of health and performance data from soccer athletes, which is open-source and publicly accessible over the Internet for coaches, players and researchers from fields such as sports science and medicine. We demonstrate a number of the applications of this dashboard on the recently released SoccerMon dataset from Norwegian elite female soccer players. SoccerDashboard can simplify the analysis of soccer datasets with complex data structures, and serve as a reference implementation for multidisciplinary studies spanning various fields, as well as increase the level of scientific dialogue between professional soccer institutions and researchers.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

• Available as PDF
• Read on any device
• Instant download
• Own it forever
• Available as EPUB and PDF
• Compact, lightweight edition
• Dispatched in 3 to 5 business days
• Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Similar content being viewed by others

Interactive Visualization of Sport Climbing Data

Visual analysis of soccer players and a team.

Health care data analysis and visualization using interactive data exploration for sportsperson

https://www.openmhealth.org/ .

https://dhis2.org/ .

https://www.acmilan.com/en/club/venues/vismara/milan-lab/ .

https://forzasys.com/pmSys.html .

Borg, G.A.: Psychophysical bases of perceived exertion. Medicine & science in sports & exercise (1982)

Google Scholar  

Carey, D.L., Crossley, K.M., Whiteley, R., Mosler, A., Ong, K.L., Crow, J., Morris, M.E.: Modeling training loads and injuries: the dangers of discretization. Med. Sci. Sports Exercise 50 (11), 2267–2276 (2018)

Article   Google Scholar  

Clarsen, B., Rønsen, O., Myklebust, G., Flørenes, T.W., Bahr, R.: The Oslo sports trauma research center questionnaire on health problems: a new approach to prospective monitoring of illness and injury in elite athletes. Br. J. Sports Med. 48 (9), 754–760 (2014)

Cross, M.J., Williams, S., Trewartha, G., Kemp, S.P., Stokes, K.A.: The influence of in-season training loads on injury risk in professional rugby union. Int. J. Sports Physiol. Perform. 11 (3), 350–355 (2016)

Drew, M.K., Finch, C.F.: The relationship between training load and injury, illness and soreness: a systematic and literature review. Sports Med. 46 (6), 861–883 (2016)

Gabbett, T.J., Domrow, N.: Relationships between training load, injury, and fitness in sub-elite collision sport athletes. J. Sports Sci. 25 (13), 1507–1519 (2007)

Gastin, P.B., Meyer, D., Robinson, D.: Perceptions of wellness to monitor adaptive responses to training and competition in elite Australian football. J. Strength Conditioning Res. 27 (9), 2518–2526 (2013)

Govus, A.D., Coutts, A., Duffield, R., Murray, A., Fullagar, H.: Relationship between pretraining subjective wellness measures, player load, and rating-of-perceived-exertion training load in American college football. Int. J. Sports Physiol. Perform. 13 (1), 95–101 (2018)

Haddad, M., Stylianides, G., Djaoui, L., Dellal, A., Chamari, K.: Session-RPE method for training load monitoring: validity, ecological usefulness, and influencing factors. Front. Neurosci. 11 , 612–613 (2017)

Hulin, B.T., Gabbett, T.J., Blanch, P., Chapman, P., Bailey, D., Orchard, J.W.: Spikes in acute workload are associated with increased injury risk in elite cricket fast bowlers. Br. J. Sports Med. 48 (8), 708–712 (2014)

Impellizzeri, F.M., Rampinini, E., Coutts, A.J., Sassi, A., Marcora, S.M., et al.: Use of RPE-based training load in soccer. Med. Sci. Sports Exerc. 36 (6), 1042–1047 (2004)

Kulakou, S., Ragab, N., Midoglu, C., Boeker, M., Johansen, D., Riegler, M.A., Halvorsen, P.: Exploration of different time series models for soccer athlete performance prediction. Eng. Proc. 18 (1), 37 (2022)

Little, R.J., Rubin, D.B.: Statistical analysis with missing data, vol. 793. John Wiley & Sons (2019)

Loturco, I., Nakamura, F., Kobal, R., Gil, S., Pivetti, B., Pereira, L., Roschel, H.: Traditional periodization versus optimum training load applied to soccer players: effects on neuromuscular abilities. Int. J. Sports Med. 37 (13), 1051–1059 (2016)

Marynowicz, J., Kikut, K., Lango, M., Horna, D., Andrzejewski, M.: Relationship between the session-RPE and external measures of training load in youth soccer training. J. Strength Conditioning Res. 34 (10), 2800–2804 (2020)

Midoglu, C., Boeker, M., Winther, A.K., Pettersen, S.A., Johansen, D., Riegler, M., Halvorsen, P., Hicks, S.: SoccerMon (2022). https://doi.org/10.17605/OSF.IO/URYZ9

Ramanathan, N., et al.: ohmage: an open mobile system for activity and experience sampling. In: 2012 6th International Conference on Pervasive Computing Technologies for Healthcare (PervasiveHealth) and Workshops, pp. 203–204 (2012). doi: 10.4108/icst.pervasivehealth.2012.248705

Rogalski, B., Dawson, B., Heasman, J., Gabbett, T.J.: Training and game loads and injury risk in elite Australian footballers. J. Sci. Med. Sport 16 (6), 499–503 (2013)

Wiik, T., Johansen, H.D., Pettersen, S.A., Baptista, I., Kupka, T., Johansen, D., Riegler, M., Halvorsen, P.: Predicting peek readiness-to-train of soccer players using long short-term memory recurrent neural networks. In: 2019 International Conference on Content-Based Multimedia Indexing (CBMI), pp. 1–6. IEEE (2019)

Download references

Author information

Authors and affiliations.

Simula Metropolitan Center for Digital Engineering (SimulaMet), Oslo, Norway

Matthias Boeker & Cise Midoglu

Oslo Metropolitan University (OsloMet), Oslomet, Norway

Matthias Boeker

Forzasys, Fornebu, Norway

Cise Midoglu

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Matthias Boeker .

Editor information

Editors and affiliations.

University of Bergen, Bergen, Norway

Duc-Tien Dang-Nguyen

Dublin City University, Dublin, Ireland

Cathal Gurrin

Radboud University Nijmegen, Nijmegen, The Netherlands

Martha Larson

Alan F. Smeaton

University of Amsterdam, Amsterdam, The Netherlands

Stevan Rudinac

National Institute of Information and Communications Technology, Tokyo, Japan

Minh-Son Dao

Department of Information Science and Media Studies, University of Bergen, Bergen, Norway

Christoph Trattner

La Trobe University, Melbourne, VIC, Australia

Phoebe Chen

Rights and permissions

Reprints and permissions

Copyright information

© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG

About this paper

Cite this paper.

Boeker, M., Midoglu, C. (2023). Soccer Athlete Data Visualization and Analysis with an Interactive Dashboard. In: Dang-Nguyen, DT., et al. MultiMedia Modeling. MMM 2023. Lecture Notes in Computer Science, vol 13833. Springer, Cham. https://doi.org/10.1007/978-3-031-27077-2_44

Download citation

DOI : https://doi.org/10.1007/978-3-031-27077-2_44

Published : 29 March 2023

Publisher Name : Springer, Cham

Print ISBN : 978-3-031-27076-5

Online ISBN : 978-3-031-27077-2

eBook Packages : Computer Science Computer Science (R0)

Share this paper

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Publish with us

Policies and ethics

• Find a journal
• Track your research

• Review Article
• Open access
• Published: 02 April 2015

Strength training in soccer with a specific focus on highly trained players

• João R Silva 1 , 2 ,
• George P Nassis 1 &
• Antonio Rebelo 2  

Sports Medicine - Open volume  1 , Article number:  17 ( 2015 ) Cite this article

96k Accesses

84 Citations

208 Altmetric

Metrics details

Data concerning the physical demands of soccer (e.g., activity pattern) suggest that a high level of performance requires well-developed neuromuscular function (NF). Proficient NF may be relevant to maintain and/or increase players’ short- (intense periods of soccer-specific activity; accelerations, decelerations, and sprinting) and long-term performance during a match and throughout the season.

This review examines the extent to which distinct modes of strength training improve soccer players’ performance, as well as the effects of concurrent strength and endurance training on the physical capacity of players.

Data sources

A selection of studies was performed in two screening phases. The first phase consisted of identifying articles through a systematic search using relevant databases, including the US National Library of Medicine (PubMed), MEDLINE, and SportDiscus. Several permutations of keywords were utilized (e.g., soccer; strength; power; muscle function), along with the additional scanning of the reference lists of relevant manuscripts. Given the wide range of this review, additional researchers were included. The second phase involved applying six selection criteria to the articles.

Results and conclusions

After the two selection phases, 24 manuscripts involving a total sample of 523 soccer players were considered. Our analysis suggests that professional players need to significantly increase their strength to obtain slight improvements in certain running-based actions (sprint and change of direction speed). Strength training induces greater performance improvements in jump actions than in running-based activities, and these achievements varied according to the motor task [e.g., greater improvements in acceleration (10 m) than in maximal speed (40 m) running movements and in non-squat jump (SJ) than in SSC-based actions (countermovement jump)]. With regard to the strength/power training methods used by soccer players, high-intensity resistance training seems to be more efficient than moderate-intensity resistance training (hypertrophic). From a training frequency perspective, two weekly sessions of strength training are sufficient to increase a player’s force production and muscle power-based actions during pre-season, with one weekly session being adequate to avoid in-season detraining. Nevertheless, to further improve performance during the competitive period, training should incorporate a higher volume of soccer-specific power-based actions that target the neuromuscular system. Combined strength/power training programs involving different movement patterns and an increased focus on soccer-specific power-based actions are preferred over traditional resistance exercises, not only due to their superior efficiency but also due to their ecological value. Strength/power training programs should incorporate a significant number of exercises targeting the efficiency of stretch-shortening-cycle activities and soccer-specific strength-based actions. Manipulation of training surfaces could constitute an important training strategy (e.g., when players are returning from an injury). In addition, given the conditional concurrent nature of the sport, concurrent high-intensity strength and high-intensity endurance training modes (HIT) may enhance a player’s overall performance capacity.

Our analysis suggests that neuromuscular training improves both physiological and physical measures associated with the high-level performance of soccer players.

Neuromuscular training improves both physiological and physical measures associated with high-level performance.

It seems that strength and power training programs should target all the force-velocity potential/spectrum of the neuromuscular system.

Due to the conditioned concurrent nature of the sport, combined strength and combined high-intensity training approaches may constitute a good training approach within a football periodized process.

Introduction

The central goal of strength/power training in a highly competitive sport is to improve the players’ specific and relevant athletic activities inherent in their sport. To achieve this outcome, different strength/power training modes with i) distinct movement patterns (traditional resistance exercises, ballistic exercises, plyometrics, weight lifting, and/or sport-specific strength-based actions), ii) different combinations of the temporal organization of strength/power training loads (e.g., microcycle and training session variations), iii) distinct loads, iv) a wide range of movement velocities, v) specific biomechanical characteristics, and vi) different training surfaces have been adopted with the final end point of achieving an improvement in players’ performance in relevant motor tasks (e.g., jumping, sprinting, and changing direction) [ 1 - 24 ].

Certain training methods combine different exercise modes (e.g., weight training, plyometric training, and sport-specific force-based actions) and allow for optimal power development and transfer to athletic activities due to both the neural and morphological adaptations typically associated with advanced training [ 25 ]. In fact, the intrinsic characteristics of soccer activity patterns (a varied range of motor actions that involve both breaking and propulsive forces as well as distinct contraction modes and velocities that require the all force-velocity potential of the neuromuscular system) that highlight the importance of the principle of specificity in strength and muscle power training cannot be understated [ 26 , 27 ].

A combination of different methods, including high-intensity strength training involving traditional resistance exercises (TRE; squats) and plyometrics [ 6 ], TRE and sprint training [ 10 ], and complex strength training (CT) [ 11 , 15 , 19 ], have all recently received considerable attention. Although some similarities exist between the previous modes of strength and power training, there are important differences. In this review, we found that complex training refers to training protocols that are comprised of the alternation of biomechanically comparable strength exercises and sport-specific drills in the same workout (e.g., six repetitions of calf extension exercise at 90% of one repetition maximum (1RM) + 5 s of rest + eight vertical jumps + 5 s of rest + three high ball headers) [ 25 ].

By focusing on more effective periodization techniques, researchers have investigated the effectiveness of different loading schemes throughout the power training phase (from high-force/low-velocity end to low-force/high-velocity end or vice versa) [ 22 ]. The training-induced effects of exercises with distinct biomechanical and technical characteristics during the plyometric-based component (e.g., purely vertically or a combination of vertically and horizontally oriented exercises [ 12 , 16 , 21 , 23 ], as well as the effects of plyometric training on different ground surfaces (grass vs. sand) [ 12 ], have both garnered significant attention. Furthermore, the adaptiveness of the functional and muscle structure of professional players (e.g., myosin heavy chain composition) to high-intensity strength training in the isokinetic contraction mode has also been investigated. However, the implementation of this analysis during the off-season resulted in lower ecological validity of these findings [ 7 ]. With regard to the search for complementary procedures and/or less stressful interventions, the effects of other methodologies (e.g., effects of electrostimulation training on semi-professional players) on physical fitness have also been investigated [ 28 ].

In general, most studies have examined the training-induced performance effects of two [ 1 , 6 , 8 , 10 , 14 , 16 , 19 ] to three [ 2 , 3 , 11 , 12 , 21 , 28 ] sessions per week. Given the multi-component requisites of soccer players’ training (e.g., endurance, speed endurance, strength, power, and agility) that coincide with the increased amount of training time, some researchers examined the short-term effect of a lower weekly volume program (one session) [ 1 , 15 , 19 ] and the effect of training-induced adaptations of different weekly training frequencies (e.g., one vs. two sessions and one session per week vs. one session every second week) on both physiological and performance parameters during pre-season [ 19 ] and throughout the in-season in well-trained soccer players [ 1 ].

Nevertheless, despite an increase in the body of evidence regarding the applicability of strength/power training programs to routine soccer training, the short-term duration of interventions (e.g., 4 to 12 weeks) [ 2 , 3 , 6 , 8 , 10 - 12 , 14 - 16 , 19 , 21 - 23 , 28 ], the wide variety of training methods, the distinct season time lines used throughout the pre-season [ 2 , 3 , 6 , 12 , 19 ] and in-season [ 8 , 14 - 16 , 21 , 24 , 28 ] periods, the different weekly training loads, and the absence of control groups make the drawing of precise conclusions very difficult. With regard to the latter aspect, it is accepted that due to the importance of winning matches, technical staff of semi-professional and professional teams are unable to implement different training scenarios based on research interests. Nevertheless, in this review, our aim is to contribute to the understanding of the present state of the art of strength/power training and concurrent training in soccer to motivate future studies.

Search strategy: databases and inclusion criteria

The selection of studies was performed in two consecutive screening phases. The first phase consisted of identifying articles through a systematic search using the US National Library of Medicine (PubMed), MEDLINE, and SportDiscus databases. Literature searches were performed from January 2013 until June 2014, and this review comprises papers from 1985 to 2014 (N 1985-2009  = 76 papers, N 2010  = 7 papers, N 2011  = 17 papers, N 2012  = 4 papers, N 2013  = 21 papers, N 2014  = 11 papers). The following keywords were used in combination: ‘elite soccer’, ‘professional soccer’, ‘first division soccer,’ ‘highly trained players,’ ‘seasonal alterations’, ‘performance analysis’, ‘soccer physiology’, ‘football’, ‘strength training’, ‘concurrent training’, ‘training transfer’, ‘neuromuscular performance,’ ‘muscular power’, ‘jump ability’, ‘sprint ability’, ‘agility’, ‘repeated sprint’, ‘intermittent endurance’. Further searching of the relevant literature was performed by using the ‘related citations’ function of PubMed and by scanning the reference lists. The second phase involved applying the selection criteria to the articles. Studies were chosen if they fulfilled the following six selection criteria: (i) the studied athletic population consisted of highly trained soccer players, ii) the players in the sample were not under 17 years of age, (iii) detailed physiological and performance tests were included, iv) the training programs applied were specified, (v) appropriate statistical analyses were used, and (vi) the article was written in the English language and published as an article in a peer-reviewed journal or a peer-review soccer-specific book edition.

Data extraction and presentation

Data related to the players’ physiological parameters (e.g., lean leg volume, body fat percentage, running economy, anaerobic threshold, maximum absolute and relative oxygen consumption and strength values, peak and mean power values, and rate force development measures) and performance parameters (e.g., soccer-specific endurance tests, maximal aerobic speed, repeated and single sprint tests, jump ability exercises, agility, and ball speed) were extracted. All data are presented as the percentage of change in the means (∆) unless otherwise specified.

Search data and study characteristics

The aim of providing players with updated data and training approaches in modern scenarios was fulfilled by 23 of the 24 papers published in the last 10 years. There were a total of 24 manuscripts fulfilling the five selection criteria, and the total sample population consisted of 523 soccer players. The distribution of players by competition level was as follows: 322 adults, 145 U-20 players, 12 U-19 players, and 44 U-18 players.

General physiological considerations of strength/power training

Strength training has become an integral component of the physical preparation for the enhancement of sports performance [ 29 ]. While strength is defined as the integrated result of several force-producing muscles performing maximally, either isometrically or dynamically during a single voluntary effort of a defined task, power is the product of force and the inverse of time, i.e., the ability to produce as much force as possible in the shortest possible time [ 9 ]. Nevertheless, strength and power are not distinct entities, as power performance is influenced by training methods that maximize both strength and stretch-shortening cycle activity (SSC) [ 30 ]. The ability of a muscle to produce force and power is determined by the interaction of biomechanical and physiological factors, such as muscle mechanics (e.g., type of muscle action) and morphological (e.g., muscle fiber type) and neural (e.g., motor unit recruitment) factors, and by the muscle environment itself (e.g., biochemical composition) [ 31 ].

The mechanisms underlying strength/power adaptations are largely associated with increases in the cross-sectional area of the muscle (hypertrophy methods) [ 32 ]. However, muscular strength increments can be observed without noticeable hypertrophy and serve as the first line of evidence for the neural involvement in the acquisition of muscular strength [ 32 ]. Thus, despite the notion that hypertrophy and neural adaptations are the basis of muscle strength development [ 33 ], their respective mechanisms of adaptation in the neuromuscular system are distinct [ 34 ]. In fact, ‘more strength’, i.e., the adaptational effect, does not necessarily imply an increase in muscle mass, as several distinct adaptations can lead to the same effect [ 33 ]. In this regard, the trainable effects of explosive/ballistic and/or heavy-resistance strength training causing enhanced force/power production have been primarily attributed to neural adaptations, such as motor unit recruitment, rate coding (frequency or rate of action potentials), synchronization, and inter-muscular coordination [ 31 , 35 , 36 ].

Physiological adaptations in soccer players

Our analysis suggests that the physiological adaptations underlining strength/power training may result in improvements in different motor tasks and performance qualities in high- and low-level players (Table  1 and Figure  1 ). In fact, independent of the players’ standard, an enhanced dynamic [ 1 - 7 , 10 , 14 , 22 , 23 ] and static maximum force production [ 4 , 5 , 28 ] and increased muscle power outputs during different physical movements can be obtained through the implementation of strength/power training routines [ 2 - 8 , 14 , 22 , 37 ]. Specifically, increases in 1RM were observed during isoinertial assessments of half-squat exercises [ 1 - 3 , 6 , 10 , 14 , 22 ], hamstring leg curls, and one-leg step-up bench exercises [ 10 ]. Additionally, in our analysis, we observed a large range of improvements in the 1RM of well-trained players after short-term intervention periods (e.g., pre-season, Figure.  1 , from 11% to 52% during the squat exercise) with average increments of approximately 21% [ 1 - 3 , 6 , 22 , 37 , 38 ]. Only Helgerud et al. [ 37 ] reported considerably larger gains in 1RM compared with other studies (11% to 26%; Table  1 ). Moreover, increments in maximal isometric voluntary contraction (MIVC) in the leg press task after CT training [ 11 ] and in knee extension strength after electrostimulation [ 28 ] and isokinetic training [ 4 , 5 ] have also been reported. Interestingly, not only were improvements in absolute force production (1RM) achieved, but an increased efficiency was also evident after allometric scaling of the results; 1RM per lean leg volume (LLV; 1RM/LLV) improved after high- and moderate-intensity modes of strength training [ 2 ], and relative force (maximum force divided by body mass) improved after complex strength training [ 11 ].

The gains in strength and d ifferent motor abilities of high-level players after 5 to 10 weeks. Squares represent the average squat jump performance [ 1 , 6 , 14 , 22 ]; rhombi represent the average countermovement jump performance [ 2 , 22 , 37 ]; triangles represent the average four bounce test performance [ 6 ]; circles represent the average 10-m sprint performance [ 2 , 22 , 37 , 38 ]; x symbols represent the average 40-m sprint performance [ 1 , 2 , 6 ]; + symbols represent the average change in direction ability [ 2 , 38 ]; and lines represent the average of all the previous motor tasks.

According to Harris et al. [ 27 ], intervention studies should use a specific isoinertial loading scheme, and test protocols should assess performance over the force-velocity continuum to gain a better understanding of the effect of load on muscular function. Moreover, neuromuscular-related qualities, such as impulse, rate of force development (RFD), and explosive strength, can better predict athletic performance; thus, the development of these approaches should be targeted [ 27 ]. The functional performance of soccer players seems to be more significantly associated with variables that are measured within the power-training load range (75% to 125% of body weight [BW] in half-squats) at which peak power (PP) is obtained (60% 1RM = 112% of BW) [ 39 ]. The PPs of highly trained soccer players were shown to occur with loads of 45% and 60% 1RM during jump- and half-squat exercises, respectively [ 22 , 39 ]. It is likely that superior improvements in power performance may be achieved by working on these optimal power training load ranges [ 22 , 39 ].

One particular muscle strength/power training adaptation involves an increase in the force-velocity relationships and the mechanical parabolic curves of power vs. velocity after high-intensity training programs, both in isoinertial [ 14 ] and isokinetic [ 4 ] exercises. Ronnestad et al. [ 6 ] and Gorostiaga et al. [ 8 ] observed increases in the force-velocity curve after high-intensity TRE and explosive-type strength training among professional and amateurs players, respectively. In the former study, the analysis of the pooled groups revealed increases in all measures of PP [ 6 ]. It seems that high-intensity strength training significantly increases performance in professional players at both the high-force end (increases in 1RM and sprint acceleration) and the high-velocity end (improvements in peak sprint velocity and four bounce test; 4BT) but only as long as the subjects perform concurrent plyometric and explosive exercises during their soccer sessions [ 6 ]. Furthermore, Los Arcos et al. (2013) recently found that professional players performing 5 weeks of pre-season and 3 weeks of in-season strength/power training increased the load at which PP was achieved during the half-squat exercise [ 11 ]. Additionally, 10 weeks of complex strength training, consisting of soccer-specific strength and skill exercises (soccer kick), improved measures of explosive strength and RFD during the isometric leg press in low-level players, with an increase in the electromyography (EMG) activity of certain muscles involved in the task also reported [ 11 ].

Adaptations in sport-specific efforts

The effectiveness of a strength/power program is evaluated by the magnitude of sport-specific improvements. Although the predominant activities during training and matches are performed at low and medium intensities, sprints, jumps, duels, and kicking, which are mainly dependent on the maximum strength and anaerobic power of the neuromuscular system, are essential skills [ 40 ]. Power and speed usually support the decisive decision-making situations in professional football, e.g., straight sprinting is the most frequent physical action in goal situations [ 41 ]. Furthermore, a high degree of stress is imposed on the neuromuscular system of players to enable them to cope with these essential force-based actions required during training and competition (e.g., accelerations and decelerations) [ 42 , 43 ].

Although not universally confirmed, there is evidence of associations between the measures of maximal (1RM) [ 44 ] and relative strength (1RM/BM) [ 45 ], as well as between certain muscle mechanical properties, such as peak torque [ 46 , 47 ] and PP [ 39 ], and the ability of soccer players to perform complex multi-joint dynamic movements, e.g., jumping and sprinting actions. Independently of a player’s level, strength-related interventions represent a powerful training stimulus by promoting adaptations in a wide range of athletic skills (e.g., jumping, Table  1 , Figures 1 , 2 and 3 and Additional file 1 : Figure S1-5) [ 2 , 3 , 6 , 8 , 10 , 12 , 14 , 15 , 19 , 21 - 23 , 48 ] and soccer-specific skills (soccer kick) [ 21 , 28 ] (Tables  1 and 2 ). Interestingly, the addition of a long-term strength/power training program to normal soccer training routines seems to result in a higher long-term increase in the physical performance of elite youth players [ 45 , 49 ]. Furthermore, to have a clear picture of the effect of strength training on physical performance, different motor tasks should be assessed; jumping, sprinting, and change of direction abilities may represent separate and independent motor abilities, and concentric and slow SSC jumping actions are shown to be relatively independent of fast SSC abilities [ 50 ].

Gains in strength and motor abilities of high level players after different training modes (5 to 10 weeks). x and dashed x symbols represent the change of direction ability performance after traditional resistance exercises programs (TRE) [ 2 ] and combined programs (COM) [ 38 ], respectively; filled and unfilled squares represent the 40-m sprint performance after TRE [ 1 , 2 ] and COM [ 6 ], respectively; + and dashed + symbols represent the 10-m sprint performance after TRE [ 2 , 37 ] and COM [ 22 , 38 ], respectively; filled and unfilled triangles represent the four bounce test performance after TRE [ 6 ] and COM [ 6 ], respectively; filled and unfilled rhombi represent the squat jump performance after TRE [ 1 , 14 ] and COM [ 6 , 22 ], respectively; and filled and unfilled circles represent the countermovement jump performance after TRE [ 2 , 37 ] and COM [ 22 ], respectively.

Percentage of improvement by training program and training session. Percentage of improvement by training program and training session after traditional resistance exercises programs (TRE), combined programs (COM), and strength/power training programs in the different motor tasks and overall functional performance (FP) of high-level players. Countermovement jump (CMJ) after TRE (CMJ-TRE) [ 2 , 20 , 37 ]; CMJ after COM (CMJ-COM) [ 22 , 23 , 38 ]; CMJ [ 2 , 20 - 23 , 37 , 38 ]; squat jump (SJ) after TRE (SJ-TRE) [ 1 , 14 ]; SJ after COM (SJ-COM) [ 6 , 19 , 22 ]; SJ [ 1 , 6 , 14 , 19 , 22 ]; 40-m sprint performance after TRE (40m-TRE) [ 1 , 2 ]; 40-m sprint performance after COM (40m-COM) [ 6 ]; 40-m sprint performance (40-m) [ 1 , 2 , 6 ]; 10-m sprint performance after TRE (10m-TRE) [ 2 , 20 , 37 ]; 10-m sprint performance after COM (10m-COM) [ 22 , 38 ]; 10-m sprint performance (10m) [ 2 , 20 - 22 , 37 , 38 ]; change of direction ability (COD) after TRE (COD-TRE) [ 2 ]; COD after COM (COD-COM) [ 38 ]; COD [ 2 , 38 ]; FP after TRE (FP-TRE) [ 1 , 2 , 6 , 14 , 20 , 37 ]; FP after COM (FP-COM) [ 6 , 19 , 22 , 23 , 38 ]; and FP [ 1 , 2 , 6 , 14 , 19 - 23 , 37 , 38 ].

Sprint ability

With regard to adaptations in sprint qualities (e.g., acceleration and maximal speed, Table  1 and Additional file 1 : Figure S1), improvements in different sprint distances (5- to 40-m distances) [ 1 , 2 , 6 , 10 - 12 , 14 , 15 , 19 , 21 , 22 , 48 , 51 ] have been reported in different levels of players. On average, highly trained players [ 1 , 2 , 6 , 22 , 37 , 38 ] need to increase their 1RM half-squat by 23.5% to achieve an approximately 2% improvement in sprint performance at 10- and 40-m distances (Figure  2 ). Excluding the study of Helgerud et al. [ 37 ], which reported significantly larger increments in strength, studies have demonstrated that lower increments in 1RM (19%) are required to achieve a similar improvement in sprint performance (1.9%) after short-term training interventions (in average, an 18% increments in 1RM resulted in a 2% average improvements in 10-m sprint performance [ 2 , 22 , 38 ] and 17% average increments in 1RM resulted in 1.6% improvements in 40-m distance time [ 1 , 2 , 6 ]). Nevertheless, improvements in sprint performance have not been entirely confirmed [ 1 , 6 , 8 , 10 , 16 , 22 , 28 ]. Notwithstanding, factors associated with the training status of various players, players’ background, and/or the characteristics of the training modes adopted should be considered as the most likely factors. For example, the sole performance of one type of plyometric exercise [ 16 ] and of electrostimulation training [ 28 ], which has an apparent lower level of specificity, may explain, at least in part, the lack of transfer of training adaptations to dynamic and complex activities, where the coordination and force production of different body muscles, as is the case of sprint performance, are essential.

Jump ability

Our analysis suggests that strength/power training induces adaptations in the jump abilities of high-level players (Table  1 and Figure  1 and Additional file 1 : Figure S2). On average, 24.4% 1RM improvements during squats result in a CMJ increase of approximately 6.8% [ 2 , 22 , 37 ]. Lower performance improvements in four bounce test (4-BT; 3.8%) were found with similar increments in 1RM (24.5%) [ 6 ], and similar improvements in SJ (6.8%) occurred with an average 1RM increase of 21.8% [ 1 , 6 , 14 , 22 ]. Curiously, the plotted data of all studies assessing the improvement in jump abilities in high-level players revealed that, on average (Figure  2 , Additional file 1 : Figure S5), a 23.5% 1RM increase may result in a 6.2% improvement in jump ability tasks after 6 to 10 weeks of strength/power training [ 1 , 2 , 6 , 14 , 22 , 37 ]. The previous results suggest that, on average, higher increments in force are needed to improve CMJ to the same extent as SJ (figure  1 ). This result may reflect the fact that the current programs were not able to increase (at the same relative rate) performance ability in the positive and negative phases of the SSC component and may explain, at least in part, the smaller improvements in sprint performance.

Improvements in the squat jump (SJ) [ 1 , 10 , 12 , 14 , 19 , 22 ], four bounce test (4BT) [ 6 ], five jump test (5-JT) [ 14 ], countermovement jump test (CMJ) [ 2 , 8 , 10 , 12 , 16 , 21 , 22 ], CMJ with free arms [ 21 ], and eccentric utilization ratio (CMJ/SJ) [ 12 ] have been observed in different players. Nevertheless, contradictions regarding improvements in SJ after plyometric [ 21 ] and in CMJ after high-intensity strength protocols performed by well-trained players can be found in the literature [ 1 , 14 ]. Additionally, no significant increases in CMJ were observed after CT involving workouts with high [ 19 ] or low loads [ 15 ] or in drop jumps from a 40-cm height (DJ 40 ) [ 10 ] following TRE and TRE plus sprint training.

Change of direction speed (COD)

According to the literature, it is difficult to discern which force/power qualities (e.g., horizontal and lateral) and technical factors influence event- or sport-specific COD ability [ 52 ]. To date, limited research has been conducted on agility/COD adaptations, with even less known about high-level athletes. Despite the limitations initially described see Introduction our results suggest that, on average, an increase of 15% in 1RM results in a 1.3% improvement in COD abilities after 5 to 6 weeks of training (Table  1 ; Figure  1 and Additional file 1 : Figure S3) [ 2 , 38 ]. Bogdanis et al. [ 2 ] observed that applying TRE-targeting hypertrophic or neural adaptations was effective in increasing COD (Table  1 , Additional file 1 : Figure S3). Nevertheless, improvements in COD performance evaluated by the 505 agility test after different plyometric techniques [ 16 ] were not found after CT [ 19 ]. Additionally, in a study by Mujika et al. [ 15 ] where players performed CT, no improvements in COD, evaluated by the agility 15-m test, were observed. The spectrum of possible factors associated with this discrepancy in results is ample and includes the players’ background and initial training status, the different training periods during which the intervention was carried out, the structure of the training intervention, game exposure, and distinct force/power qualities and technical factors that influence event- or sport-specific COD. For example, the study of Maio Alves et al. [ 19 ] was implemented during pre-season, and the research of Thomas et al. [ 16 ] was carried out during in-season. Consequently, the accumulated effect of COD actions performed during training sessions and games may influence these results [ 46 , 53 ]. Although the players are from the same age groups, the differences in the competitive levels of the players from previous studies should not be ignored. Moreover, the lack of improvements in COD after in-season CT that are reported by Mujika et al. [ 15 ] may be related to the fact that only six sessions were performed in a 7-week period. As will be further analyzed (‘Training efficiency’), this fact, among others, may suggest that higher training volumes may be necessary to induce adaptations in COD.

Sport-specific skills

One of the most important indicators of a successful soccer kick is the speed of the ball. Studies involving amateur players observed that CT [ 11 ] and electrostimulation training [ 28 ] increase ball speed with [ 11 , 28 ] and without (Table  1 ) run up [ 28 ]. Nevertheless, these improvements were examined in lower standard players. Moreover, elite U-19 players performing plyometric training increased ball speed with the dominant and non-dominant leg [ 21 ]. Other studies involving elite players performing different modes of strength training (isokinetic strength training or functional training) did not report improvements in ball speed [ 4 , 5 ]. Nevertheless, in studies performed during the off-season period, training stimulus consists of the exercise mode of the experimental designs and no other types of soccer routines are undertaken. Thus, the results should be analyzed with caution as the scenarios for training transfer to occur during this period are constricted (off-season); the increases in certain strength parameters were not reflected in positive transference to consecutive gains in ball speed.

Comparing different training variables in strength/power interventions in soccer

The multi-factorial constructs of soccer performance (technical, tactical, and physical performance) and their associated components bring a higher complexity to the designing of the training process. In fact, professionals involved in the preparation of soccer teams have to reflect on several questions associated with the manipulation of the individual variables that affect each of these relevant constructs and how they can affect each other. With regard to physical performance, several potential questions arise: What are the most beneficial movement patterns and type of training? How many sessions do athletes need to improve and maintain the performance outcome? Does ground surface have an effect on adaptations? We will analyze these and other relevant questions in the following sections.

Force production and movement pattern specificity: traditional resistance exercises vs. combined programs

Our analysis suggests that the activity patterns of applied exercises may influence performance outcomes (Figures  2 and 3 and Additional file 1 : Figure S4 to S5). Therefore, we compared programs involving mainly traditional resistance exercises (TREs) with programs that combine different activity patterns during the training intervention (COM; programs including TRE and ballistic exercises, plyometrics, weight lifting, body weight exercises, and/or sprint training during training cycles). Despite the fact that some limitations can be ruled out from this type of analysis (e.g., differences in session and weekly training volumes and load, the density of different intrinsic activity patterns, and the 1RM percentage used during the loaded exercises), we believe that it will aid in challenging research designs in this field.

Effects on sprint performance

On average, despite TRE resulting in superior strength gains compared with COM, greater performance improvements in the 10-m sprint are observed after COM (TRE = in average, 26.8% increments in 1RM resulted in 1.93% average improvements in 10-m sprint [ 2 , 37 ]; COM = in average, 19.9% increments in 1RM resulted in 2.4% average improvements in 10-m sprint [ 2 , 22 , 38 ]; Figure  2 and Additional file 1 : Figure S5). However, our analysis suggests the opposite with regard to 40-m sprint performance (TRE = in average, 15.8% increments in 1RM resulted in 1.9% average improvements in 40-m time [ 1 , 2 ] COM = in average, 23% increments in 1RM resulted in 1.1% average improvements in 40-m sprint time [ 6 ]). Nevertheless, all pooled data suggest that despite the TRE result of greater increases in 1RM (26%) than COM (21%), this may not translate into superior improvements in the sprint performance of high-level players (1.9% TRE vs. 2.1% COM; Additional file 1 : Figure S4).

Effects on jump ability

By performing the same analysis for jump ability exercises (Figure  2 and Additional file 1 : Figure S5), we found that there is a tendency toward greater strength increases after TRE (in average, 26.8% increments in 1RM resulted in 6.8% average improvements in CMJ; in average, 22% increments in 1RM resulted in 6.7% average enhancement in SJ; in average, 25% increments in 1RM resulted in 6% average improvements in 4BT) that are not translated into superior performance gains compared with the results observed following COM (in average, 21% increments of 1RM resulted in 6.8% average improvements in CMJ; in average, 22% increments in 1RM resulted in 6.9% average enhancements in SJ; in average, 22% increments of 1RM resulted in 6.4% average improvements in 4BT). In fact, all pooled data show that greater improvements in jump ability may be obtained with lower strength increases after COM than TRE only (Additional file 1 : Figure S5; in average, 21.6% increments in 1RM resulted in 6.4% average improvements in jump ability and a 25% average increments in 1RM resulted in 6% average improvements in jump ability, respectively). This higher efficacy of transfer of strength gains to performance improvements after COM seems to be more evident in SSC jump ability (CMJ). Taking into consideration, among other factors, the described associations between physiological and mechanical characteristics (e.g., post-activation potentiation and peak torque) and CMJ and running-based actions in professional players [ 44 , 46 , 54 ], this fact may suggest that COM may represent a superior method for improving sport-specific actions compared with TRE alone. Additional studies on this topic are necessary.

Effects on COD ability

Given the scarcity of literature assessing the effect of COD training modes and the reported small to moderate associations between strength and power variables with COD performance and different characteristics (e.g., test duration, COD number, and primary application of force throughout the test) of the agility tests commonly used to evaluate COD [ 52 ], conclusions should be drawn with caution. In fact, within programs involving only TRE, as will be discussed later in this review (‘ Manipulation of loading schemes ’), it seems that manipulating different mechano-biological descriptors of strength/power stimuli may influence performance adaptations in COD actions [ 2 ]. Nevertheless, our analysis shows that, on average, lower strength increases after TRE [ 2 ] produce greater performance improvements in the agility t -test than after COM [ 38 ] (in average, 14.2% increments in 1RM resulted in 1.7% average improvements in t -test and a 19.9% average increment in 1RM resulted in 1% average improvement in t -test, respectively; Figure  2 ).

Two studies are particularly relevant with regard to this topic: TRE vs. TRE plus plyometrics [ 6 ] and TRE vs. TRE plus sprint training [ 10 ]. In the study of Ronnestad et al. [ 6 ], although no significant differences between groups were observed, the group of players who utilized combined approaches broadly improved their performance. Additionally, Kotzamanidis et al. [ 10 ] observed that the jump and sprint performance of low-level players only improved in the combined program approach. Thus, it seems that combining heavy and light load training schemes may be an effective method for improving muscular function and may be particularly useful when force application is required in a wide range of functional tasks [ 27 ].

Training efficiency

To estimate the improvement in the different motor tasks and in overall functional performance, as well as the efficiency (efficiency = percentage of improvement/number of training sessions) of strength/power interventions and the effects of the different types of programs (TRE vs. COM) on specific motor tasks and functional performance, we performed an analysis involving all studies in highly trained players where performance outcomes were reported despite no references to changes in force production (Figure  3 ). Despite the limitations already highlighted, our analysis suggests that even though TRE slightly increases overall functional performance, the efficiency (gains by session) is lower than in COM modes. These uncertainties make this research topic particularly crucial. In summary, considering the high demands of high-level competition, the increase in different motor tasks (1.3% to 7.2%) and overall functional performance (4%) observed in highly trained players following strength/power training programs makes strength/power programs an essential training component. In general, it seems that strength/power training induces greater improvements in jump abilities than in running-based activities. Moreover, combining resistance- and speed-training or plyometric- and soccer-specific strength programs in the same session seems to be more effective than the resistance-training program alone [ 6 , 10 , 48 ].

Manipulation of loading schemes

Bogdanis et al. [ 2 , 3 ] analyzed the effects of high-repetition/moderate-load (hypertrophy) and low-repetition/high-load (neural adaptations) programs on anthropometric, neuromuscular, and endurance performance. These last studies [ 2 , 3 ] and others [ 4 , 5 , 23 ] suggest that the manipulation of different mechano-biological descriptors of strength/power stimuli (e.g., load magnitude, number of repetitions) is associated with different physiological and performance adaptations in highly trained soccer players. The hypertrophic mode was associated with increases in lower limb muscle mass, while the neural mode was more effective in improving 1RM/LLV, sprint, and COD performance [ 2 ]. In another study, Bogdanis et al. [ 3 ] found that even though both groups (hypertrophic group vs. neural group) improved the total work performed during a repeated cycle ergometer sprint test (RST; 10 × 6-s sprint with 24-s passive recovery), the neural mode group had a significantly greater improvement in work capacity during the second half (sprint 6 to 10; 8.9% ± 2.6%) compared with the first half of RST (sprint 1 to 5; 3.2% ± 1.7%). These results suggest that the neural mode confers a higher fatigue resistance during RST [ 3 ]. In addition, the mean power output expressed per lean leg volume (MPO/LLV) was better maintained during the last six sprint post-training only in the neural group, and there was no change in MPO/LLV in the hypertrophic group in the RST [ 3 ]. These results suggest, at least in part, a better efficacy of neural-based programs in high-level players [ 2 , 3 ] that could be linked to several adaptive mechanisms that are not associated with increases in muscle volume. However, the most likely adaptations are at the neuro-physiological level, i.e., changes in the pattern of motor unit recruitment and increases in rate coding [ 2 , 32 ].

Other researchers observed that physiological and performance outcomes can be independent of the kinetics of the power loading scheme used (from the high-force/low-velocity end to the low-force/high-velocity end and vice versa) because the loading scheme components spanned the optimal power training spectrum [ 22 ].

Contraction modes

The analysis of the impact of high- vs. low-intensity isokinetic strength vs. functional strength showed that professional players who performed a high-load, low angular velocity program had a higher improvement in maximal isometric and isokinetic strength and in PP at different knee angles and velocities [ 4 , 5 ]. Although the increases in dynamic muscle strength were generally associated with the specific velocities used in the training programs, the high-load/low-velocity group also exhibited improvements in muscle force and power at high knee extension velocities [ 4 , 5 ]. Although several explanations can be offered to clarify the greater adaptations associated with a wide range of velocities observed after the high-load/low-velocity strength training program, the most likely explanation is the occurrence of changes in neural and morphological factors associated with this type of training (e.g., increases in RFD, muscle mass, and/or fiber pennation angle).

Training frequency

As previously mentioned, high-level soccer players are usually involved in weekly matches of national leagues and are often involved in international commitments, thus limiting the time available for fitness training. Maio Alves et al. [ 19 ] found that different weekly volumes (two vs. one session per week) of complex training performed by high-level junior players resulted in similar improvements in sprint, jump, and COD ability. Ronnestad et al. [ 1 ] observed that one high-intensity strength training session per week during the first 12 weeks of the in-season period represented a sufficient training stimulus for maintaining the pre-season (two sessions per week for 10 weeks) gains in strength, jump, and sprint performance of professional players. However, a lower weekly in-season volume (one session every two weeks) only prevented detraining in jump performance [ 1 ]. Accordingly, a recent study [ 48 ] involving a larger sample of players showed that professional teams subject to distinct weekly strength training stress (all performed one resistance strength session a week) exhibit higher neuromuscular performance in the middle of the season than at the start of the season. Nevertheless, only the team that performed a higher number of sessions targeting the neuromuscular system showed improved neuromuscular performance during the second phase of the season. Despite the distinct individual variables that constituted the weekly resistance training session performed by the teams (e.g., percentage of 1RM, number of repetitions and exercises), differences in strength/power training stress were mainly due to the higher employed volume of both soccer-specific strength and sprint sessions [ 48 ]. This result again established the important role of the specificity of the training stimulus. Given the important role of circulating levels of androgens in strength and power performance, it is relevant to mention that only the high neuromuscular training scheme positively affected the circulation and activation (increase in 3a Diol G) of the androgen pool (total testosterone) [ 48 ].

However, Mujika et al. [ 15 ] observed that a low volume of combined forms of strength/power training is more effective in improving sprint performance (15-m sprint time) than the sole performance of lower volumes of sprint training in elite U-19 players.

Manipulation of biomechanical components of plyometric-based exercises

Performance outcomes may also be influenced by the biomechanical nature of the exercises employed in a single or combined program. Los Arcos et al. [ 23 ] observed that weight training plus plyometric and functional exercises involving vertically and horizontally oriented movements were more effective in enhancing the CMJ performance of highly trained players than exercises involving purely vertically oriented movements. Nevertheless, both groups improved their PP and showed small, although non-significant, improvements in 5- and 15-m sprint performance [ 23 ]. In contrast, Thomas et al. [ 16 ] examined that both plyometric training involving drop jumps or CMJs were effective in improving the jump (CMJ) and COD ability (505 agility test) of semi-professional players, regardless of the lack of change in short sprint distances. It is important to highlight that although no between-group differences were reported, the improvements in COD ability were twofold greater in the CMJ group. Nevertheless, given the age group of the players (U-18), it is important to be cautious in extrapolating these findings to professional adult players.

Training surface

There is also evidence that the ground surface used during plyometrics (sand vs. grass) may influence adaptations [ 12 ]. Impellizzeri et al. [ 12 ] observed that performing plyometrics on grass produced greater effects in CMJ and in the eccentric utilization ratio CMJ/SJ than when performed on sand. However, a trend toward higher adaptations was observed in SJ when the training program was performed on sand (Table  1 ). Additionally, sand was found to induce lower levels of muscle soreness compared with grass [ 12 ]. The fatigue development and recovery kinetics during and after a game have been well characterized in recent years. A reduction in the players’ ability to produce force toward the end of the match and in the match recovery period, an increase in some indirect markers of muscle damage, and longer periods of post-match muscle soreness have all been described [ 55 - 68 ]. In light of these findings, it may be expected that sandy surfaces may be a good alternative for the execution of plyometric programs during periods of high-volume, high-intensity, or high-frequency training (e.g., pre-season) and when athletes are recovering from injury and trying to regain physical capacity. In fact, in addition to improving neuromuscular capabilities, sand has been shown to produce lower levels of muscle soreness compared with grass [ 12 ]. Accordingly, compared with natural grass or artificial turf, the performance of dynamic powerful actions on sand, despite the known higher energy expenditures and metabolic power values, results in smaller impact shocks and limited stretching of the involved muscles [ 69 ].

Interference between concurrent strength and endurance training

Concurrent training involves the incorporation of both resistance and endurance exercises in a designed, periodized training regime [ 70 ]. The current dogma is that muscle adaptations to RE are blunted when combined with endurance [ 71 ], resulting in lower strength and power gains than those achieved by resistance exercise alone. When the modes of strength and endurance training focus on the same location of adaptation (e.g., peripheral adaptations), the muscle is required to adapt in distinctly different physiological ways [ 72 ]. However, when the modes of strength/power and endurance training are at opposite ends of the biomechanical and neuro-coordinative spectrum, the anatomical and performance adaptations may be reduced, and the accuracy of the intended movement, fluidity, and elegance that characterize excellence may be compromised. In fact, it is the entire spectrum of characteristics (e.g., metabolic and neuro-coordinative) of the upstream stimulus (resistance vs. endurance exercise; RE vs. E) that determines the downstream events necessary for training adaptations to occur. The range of factors that may be associated with the interference phenomenon or the incapability of achieving/maintaining higher levels of strength/power during concurrent strength and endurance training is ample and spans from excessive fatigue or increments in catabolic environments to differences in motor unit recruitment patterns, possible shifts in fiber type, and conflicts with the direction of adaptation pathways required by the muscle [ 34 , 70 , 72 , 73 ].

Molecular events

RE stimulates a cascade of events leading to the induction or inhibition of muscle atrophy [ 74 ]. From a molecular standpoint, these adaptations result from the downstream events promoted by the phosphatidylinositol 3-kinase/protein kinase B/mammalian target of rapamycin (PI3-k/Akt/mTOR) pathway [ 74 , 75 ]. However, three kinases [p38 mitogen-activated protein kinase (MAPK), AMP-activated protein kinase (AMPK), and calmodulin-dependent protein kinase] are particularly relevant in the signaling pathways that mediate skeletal muscle adaptations to endurance-based training [ 75 , 76 ].

A few studies highlight the notion that both translation efficiency and protein synthesis may be compromised due to the incompatibility of the two different intracellular signaling networks, i.e., activation of AMPK during endurance exercise impairs muscle growth by inhibiting mTOR [ 74 , 75 ]. Nevertheless, other studies revealed that endurance performed after RE did not compromise the signaling pathways of RE (mTORC1-S6K1) [ 71 ] and may amplify the adaptive response of mitochondrial biogenesis [ 76 ]. Moreover, the translational capacity for protein synthesis can be reinforced rather than compromised when aerobic exercise precedes RE and molecular events are not compromised; mTOR and P70S6K shown greater phosphorylation in response to concurrent aerobic exercise compared with RE alone [ 77 ]. Furthermore, chronic concurrent aerobic exercise and RE may increase aerobic capacity and promote a greater increase in muscle size than RE alone [ 78 ]. Nevertheless, taking into account the complexity and the several molecular interactions that constitute the cascade of events associated with resistance and endurance exercise, conclusions should be drawn with caution. Additionally, studies have been performed primarily in healthy adults (physically active college students, moderately trained and recreationally active subjects) and not high-level athletes; although not universally confirmed, athletes with more extensive training backgrounds may have distinct phenotypes [ 79 - 81 ] and genotypes than normally active subjects [ 82 ]. Moreover, to the authors’ best knowledge, there is no research concerning how the distinct genotypes that can be found within a high-level group of athletes [ 82 - 84 ] may influence the individual responses to concurrent training.

Methodological considerations

Given the divergent physiological nature of strength and endurance training [ 34 ], the methodology applied, the volume and frequency of training, and the target goal all play key roles in increasing the degree of compatibility between these two key physical fitness determinants [ 34 , 72 ]. Slow long-duration sustained aerobic conditioning (SLDC) has been shown to be potentially detrimental to the overall performance of athletes involved in power sports and, for example, may have a negative impact on strength and power development [ 85 ]. Excessive training volumes may contribute to high metabolic stress, leading to high levels of substrate depletion and catabolic states (e.g., increased cortisol responses) [ 85 ]. Furthermore, SLDC may compromise recovery and regeneration, leading to a progression in the overtraining continuum [ 85 ]. Moreover, the high levels of oxidative stress (e.g., damaging proteins, lipids, and DNA) that are associated with high-volume training may increase reactive oxygen species (ROS) production to a level that overcomes the positive adaptations that may be triggered by ROS, i.e., there is a range in which ROS may represent an optimal redox state for greater performance, as with force production capacity [ 86 ]. Additionally, these previous factors associated with SLDC that limit force production may compromise skill acquisition by reducing the quality of execution (e.g., the technical ability of force application) and, thus, motor learning [ 85 ]. It is reasonable to consider that there may be certain mechanisms associated with the combination of training modalities that produce positive improvements and are additive in nature [ 87 ].

A low-volume, high-intensity approach, such as sprint interval training, may favor an anabolic environment (e.g., growth hormone, insulin-like growth factor-I, IGF binding protein-3, and testosterone) [ 88 - 92 ], maintain a muscle fiber phenotype associated with strength and power capabilities [ 93 ], and increase endurance and neuromuscular-related outcomes [ 94 - 96 ]. In fact, HIT and/or combined forms of HIT seem to promote adaptations in skeletal muscle and improvements in laboratory and field endurance-related parameters that are comparable to the effects of high-volume endurance training [ 94 , 97 - 101 ] and may improve muscle power-based actions [ 94 , 102 ]. Interestingly, the type of previously observed hormonal responses to HIT (e.g., sprint interval training) [ 88 - 92 ] constitutes one of the paradigms of resistance exercise biology, namely, an increase in cellular signaling pathways as well as satellite cell activation that contributes to an increase in translation and transcription processes associated with protein synthesis [ 74 ]. In this regard, supramaximal interval training is shown to be superior to high-intensity interval training for concurrent improvements in endurance, sprint, and repeated sprint performance in physically active individuals [ 103 ].

Does the magnitude of neuromuscular involvement during training sessions reduce possible incompatibilities associated with concurrent training? Are the biomechanical and neuro-coordinative demands (e.g., accelerations/decelerations impacting mechanical load and neuromuscular demands) of different training modes with similar physiological responses the same (e.g., 4 × 4-min interval running with 2-min rest vs. 4 × 4-min SSG with 2-min rest vs. 4 × 4-min intermittent situational drill with 2-min rest)? It is possible that, from a biomechanical and neuromuscular standpoint, more specific training methods to develop strength/power and endurance performance with higher biomechanical and neuromuscular demands may improve both adaptations and performance outcomes, as well as reduce the negative effect of this interference from a molecular point of view; human-based studies to date are far from agreement regarding the molecular interference after acute concurrent exercise [ 70 ]. In fact, strength/power and HIT are characterized by brief intermittent bouts of intense muscle contractions. Questions related to training transfer should be observed with greater attention when extrapolating the applicability of concurrent training to sport-specific settings. In fact, several factors can influence the transfer of strength training in endurance performance and the impact of endurance workloads on strength and power performances [ 104 ].

Soccer: a concurrent modality

A soccer player’s performance is intimately associated with the efficiency of different energy-related systems [ 105 - 107 ]. During the season, players perform intense programs with multiple goals of increasing strength, power, speed, speed endurance, agility, aerobic fitness, and game skills [ 108 ]. In fact, despite the predominant activity patterns of the game being aerobic in nature, the most deterministic factors of match outcome depend on anaerobic mechanisms [ 41 ]. It is common sense that the most intense match periods and worst-case match scenarios are associated with periods of high mechanical and metabolic stress. In fact, recently developed techniques of match analysis provide a body of evidence that supports the belief that neuromuscular demands of training and competition are higher than initially suspected (e.g., accelerations/decelerations) [ 42 , 43 , 109 ] and give further support to the viewpoint that strength/power-related qualities are crucial for high-level performance.

There is a belief that by stressing the neuromuscular system, adaptive mechanisms that are neurological, morphological, and biomechanical in nature will be triggered, thus increasing the player’s neuromuscular performance and providing him/her with a superior short- and long-term endurance capacity [ 17 , 110 - 113 ]. In this regard, associations between neuromuscular qualities (e.g., CMJ peak power) and intermittent endurance exercise [ 114 ] and repeated sprint ability performance [ 115 ] have also been observed. Moreover, there has been evidence supporting the association between team success and jump abilities (e.g., CMJ and SJ) [ 116 ]. Additionally, starter players demonstrate higher strength [ 108 ] and power performance capabilities than non-starters [ 117 ], and greater neuromuscular capabilities have been associated with game-related physical parameters and lower fatigue development during matches [ 118 ]. Moreover, Meister et al. [ 119 ] observed that after a match congestion period, players with a higher exposure time show better scores in certain neuromuscular parameters (CMJ, drop jump height, and drop jump contact) than players with a lower exposure time, although this result is not significant. Interestingly, recent reports revealed that neuromuscular-based actions, such as sprinting, have improved more in recent years than physiological endurance parameters. Professional players tested during the 2006 to 2012 seasons actually had a 3.2% lower VO 2 max than those tested during 2000 to 2006 [ 120 , 121 ]. Although with the obvious limitations and the universal consensus of the importance of aerobic fitness in soccer, these observations suggest that anaerobic power is ‘ stealing space ’ from aerobic power with regard to the constructs relevant in soccer performance. All of these previous facts highlight the role of neuromuscular exercise during soccer training and suggest that soccer routines should be performed concurrently as they are concurrent by nature. In fact, the physiological systems associated with endurance fitness development and maintenance are generally largely targeted in any match competition, friendly game, tactical exercise, circuit technical drills that often involve frequent displacements, and/or small side game exercises performed during a 90-min soccer competition/training session [ 106 , 122 , 123 ].

Physiological and performance adaptations

The summary of changes in physiological and functional parameters resulting from concurrent strength and endurance training are presented in Table  2 . Wong et al. [ 20 ] observed that 8 weeks of pre-season high-intensity strength training and SE resulted in a significant improvement in endurance markers, soccer-specific endurance (SSE), and soccer-specific neuromuscular (SSN) parameters. Helgerud et al. found that 8 weeks of other modes of HIT (aerobic high-intensity training) and high-intensity strength training during the preseason of non-elite [ 51 ] and elite [ 37 ] football players improved VO 2 max (8.6% and 8.9%), running economy (3.5% and 4.7%), and 1RM during half-squat strength exercise (52%), respectively. Moreover, the 10- and 20-m sprint performance (3.2% and 1.6%, respectively) and CMJ (5.2%) of elite players also improved [ 37 ]. These strength improvements occurred with minor increases in body mass (average 1%) and a substantial increase in relative strength [ 37 ]. More recently, McGawley et al. [ 38 ] found that a high-frequency program (three times a week) of concurrent high-intensity running-based training with strength/power-based training in the same session resulted in a positive training effect on all evaluated measures, ranging from flexibility, anthropometric, endurance, and neuromuscular-related parameters (Table  2 ). Moreover, these results suggested that the order of completion of the program, E + RE or RE + E, did not influence the performance adaptations. These results [ 38 ] and others [ 2 , 37 ] may support, at least in part, the better compatibility between high-intensity modes of strength and endurance training.

It is reasonable to assume that the players in the studies examining the effects of strength training programs (Table  1 ) had performed training with significantly high weekly endurance-based loads (e.g., pre-season). In this regard, Bogdanis et al. [ 3 ], when examining the strength training effects of the hypertrophic and neural modes in professional soccer players during pre-season, reported that the weekly cycle also involved a considerable amount of interval training and small-sided games, which have been described as effective methodologies targeting endurance fitness and SSE development (for a review, see [ 95 , 122 ]). The authors [ 3 ] observed that both aerobic fitness parameters (e.g., VO 2 max and MAS) and SSE, evaluated by the Yo-Yo intermittent endurance test and Hoff’s dribbling track test, respectively, were significantly improved in both groups (Table  1 ). Furthermore, other researchers [ 23 ] found that strength/power training performed in parallel with endurance training resulted in improvements in the individual anaerobic threshold and muscle/power parameters. Additionally, the performance of explosive-type strength training with routine soccer training did not interfere with the aerobic capacity of amateur young players [ 8 ], e.g., sub-maximal blood lactate values. These findings suggest that performing concurrent strength/power training and routine soccer training is advisable because, in addition to an increase in neuromuscular performance and the anabolic environment, this training did not interfere with the development of aerobic capacity [ 8 ]. Nevertheless, the question of whether this compatibility is related to the type of endurance and strength performed is highlighted in the distinct between-group results presented in the study of Bogdanis et al. [ 3 ], e.g., point ‘ Manipulation of loading schemes ’, where only the neural group significantly improved with respect to running economy and a trend toward a better performance in the YYIE2 in the neural group than in the hypertrophic group was reported.

In another study [ 13 ], semi-professional male soccer players performed both endurance and strength sessions as part of the annual periodization (four cycles of 12 weeks). This type of periodization was effective in improving both the endurance performance (Probst test) and SSN parameters, e.g., CMJ. These results suggested that no adaptation conflicts occur when one or two sessions of strength/power and endurance are simultaneously combined during a soccer training cycle (endurance block composed of two endurance training sessions and one strength training session and vice versa).

Additionally, Lopez-Segovia et al. [ 18 ] examined training adaptations in elite U-19 players during a 4-month period. The training program consisted of four sessions per week, targeting the improvement of player’s aerobic performance. Training was complemented with one or two specific strength training sessions per week performed at the start of the training session. This type of periodization improved loaded CMJ performance and the speed of movement in full squats, with loads ranging from 20 to 40 kg. Nevertheless, significant decrements in different sprint abilities were found. According to the researchers, the lack of improvement in the former sprint variables was attributed to the high volume of aerobic work performed. Nevertheless, an increase in MAS (3.2%) was observed after the intervention period [ 18 ].

Conclusions

Our analysis suggests that, independent of the methodology applied (Table  1 ) and the form of concurrent endurance and strength/power training (Table  2 ), pre-season training resulted in an improvement in physiological and soccer-specific and non-specific performance parameters. The large responsiveness to training may be associated with the fact that most of the studies were conducted during an early stage of pre-season, with off-season detraining negatively affecting several physical attributes, such as anthropometric characteristics (e.g., decreases in LBM and increases in BF) [ 124 - 126 ], endurance-related markers [ 53 , 101 , 126 , 127 ], soccer-specific endurance [ 101 , 128 ], and neuromuscular parameters [ 126 , 129 ]. With this in mind, the overall conclusion of the analyzed literature is that the addition of strength/power training programs to routine soccer training favors a more integral physical fitness development of the player. The associated improvements in physiological (e.g., 1RM/LLV, PP) and performance (e.g., jump, sprint, COD) parameters may, at least in part, increase a player’s ability to cope with training and competition demands. Our analysis suggests that high-intensity strength training (HIST) may be a more efficient method than moderate-intensity methods (hypertrophic). In addition, the compatibility between strength and endurance training may be greater when high-intensity or explosive strength training is combined with high-intensity endurance training to favor a more soccer-specific phenotype.

One of the most sensitive periods of training implementation is the in-season period. As the match is the most important part of the soccer-training schedule, technical staff often view the in-season periodization with particular prudence. They want to maintain or even increase the pre-season gains obtained throughout the short pre-season period (5 to 7 weeks). However, they face the constant dilemma of determining the proper dose/response that allows for the cycle of training-recovering/competing-recovering to be effective; a high volume of training and/or competition interspersed by insufficient recovery favors fatigue development [ 130 ], resulting in a transition from a functional to a non-functional overreaching state or, in more severe cases, an overtraining state [ 131 , 132 ]. Unfortunately, studies implemented during in-season are scarce [ 1 , 8 , 13 - 16 , 18 , 21 , 24 , 28 , 48 ]; seven were conducted with U-19 players, and only four were conducted with adult soccer players [ 1 , 13 , 28 , 48 ]. Our analysis suggests that two weekly sessions allow for highly trained players to obtain significant performance enhancements and that one session a week is sufficient to avoid in-season detraining. It may be possible that, in parallel with a higher volume of neuromuscular training (soccer-specific strength/power-based efforts), further in-season improvements could be observed. Moreover, manipulations of the training surface could constitute an important strategy (e.g., players returning from injury and the management of biochemical and perceptual disturbances).

We found that the results of high-force increments vs. low-performance enhancements and the respective efficiency of the programs (jump vs. running-based actions and non-SSC abilities (SJ) vs. SSC-based actions (e.g., CMJ)) suggest that current approaches may overlook some essential aspects required to achieve an increase in a player’s performance capacities. According to Komi [ 133 ], an effective SSC is obtained with ‘a well-timed pre-activation of the muscle(s) before the eccentric phase, a short and fast eccentric phase, and an immediate transition (short delay) between stretch (eccentric) and shortening (concentric phase).’ The observed increments in force production will most likely occur to a greater extent in the positive phase of the SSC. We suggest that to achieve greater improvements, weight training should be combined with more soccer-specific strength exercises (e.g., the player’s ability to use strength and power effectively and consistently [ 134 ], allowing for the application of force/power in a larger range of planes (horizontal) and specific angles). Therefore, a conditioning method such as Speed, Agility and Quickness (SAQ) may be useful, as it incorporates plyometric and soccer-specific strength exercises and can, therefore, constitute a good conditioning tool for this type of outcome (acting on the entire spectrum of the SSC and on the transition from eccentric to concentric movements; it should be kept in mind that plyometric training is a technique demonstrated to increase musculo-tendinous stiffness, which can optimize power output in explosive movements) [ 135 ]. The greater ecological validity of COM approaches make combined methods a preferred training strategy for strength training in soccer; targeting the intra- and inter-muscular aspects of athletic performance should occur in parallel and begin at the start of the preparation period. In fact, hypertrophy and general power exercises can enhance sports performance, but optimal transfer from football-specific activities also requires football-specific exercise programs [ 29 ] in which the biomechanical and neuro-coordinative patterns of sport-specific motor tasks are taxed.

In summary, the analyzed literature suggests that the training of neuromuscular function and its combination with soccer-specific endurance results in improvements in non-specific (e.g., anthropometric characteristics, relative strength, and VO 2 max) and soccer-specific endurance and neuromuscular parameters (e.g., YYIER, RSA, and sprint).

Ronnestad BR, Nymark BS, Raastad T. Effects of in-season strength maintenance training frequency in professional soccer players. J Strength Cond Res. 2011;25(10):2653–60.

Article   PubMed   Google Scholar  

Bogdanis GC, Papaspyrou A, Souglis A, Theos A, Sotiropoulos A, Maridaki M. Effects of hypertrophy and a maximal strength training programme on speed, force and power of soccer players. In: Reilly T, Korkusuz F, editors. Science and Football VI. The proceedings of the sixth world congress on science and football. New York: Routledge; 2009. p. 290–5.

Google Scholar  

Bogdanis GC, Papaspyrou A, Souglis AG, Theos A, Sotiropoulos A, Maridaki M. Effects of Two Different Half-Squat Training Programs on Fatigue During Repeated Cycling Sprints in Soccer Players, Journal of strength and conditioning research / National Strength & Conditioning Association. 2011. May 12.

Aagaard P, Trolle M, Simonsen E, Bangsbo J, Klausen K. High speed knee extension capacity of soccer players after different kinds of strength trainning. In: Reilly T, Clarys J, Stibbe A, editors. Science and Footbal II. London: F & FN Spon; 1993. p. 92–4.

Trolle M, Aagaard P, Simonsen E, Bangsbo J, Klausen K. Effects of strength trainning on kicking performance in soccer. In: Reilly T, Clarys J, Stibbe A, editors. Science and Footbal II. London: F & FN Spon; 1993. p. 95–7.

Ronnestad BR, Kvamme NH, Sunde A, Raastad T. Short-term effects of strength and plyometric training on sprint and jump performance in professional soccer players. J Strength Cond Res. 2008;22(3):773–80.

Andersen JL, Klitgaard H, Bangsbo J, Saltin B. Myosin heavy chain isoforms in single fibres from m. vastus lateralis of soccer players: effects of strength-training. Acta Physiol Scand. 1994;150(1):21–6.

Article   CAS   PubMed   Google Scholar  

Gorostiaga EM, Izquierdo M, Ruesta M, Iribarren J, Gonzalez-Badillo JJ, Ibanez J. Strength training effects on physical performance and serum hormones in young soccer players. Eur J Appl Physiol. 2004;91(5–6):698–707.

Hoff J, Helgerud J. Endurance and strength training for soccer players: physiological considerations. Sports Med. 2004;34(3):165–80.

Kotzamanidis C, Chatzopoulos D, Michailidis C, Papaiakovou G, Patikas D. The effect of a combined high-intensity strength and speed training program on the running and jumping ability of soccer players. J Strength Cond Res. 2005;19(2):369–75.

PubMed   Google Scholar  

Manolopoulos E, Papadopoulos C, Kellis E. Effects of combined strength and kick coordination training on soccer kick biomechanics in amateur players. Scand J Med Sci Sports. 2006;16(2):102–10.

Impellizzeri FM, Rampinini E, Castagna C, Martino F, Fiorini S, Wisloff U. Effect of plyometric training on sand versus grass on muscle soreness and jumping and sprinting ability in soccer players. Br J Sports Med. 2008;42(1):42–6.

Nunez VM, Da Silva-Grigoletto ME, Castillo EF, Poblador MS, Lancho JL. Effects of training exercises for the development of strength and endurance in soccer. J Strength Cond Res. 2008;22(2):518–24.

Chelly MS, Fathloun M, Cherif N, Ben Amar M, Tabka Z, Van Praagh E. Effects of a back squat training program on leg power, jump, and sprint performances in junior soccer players. J Strength Cond Res. 2009;23(8):2241–9.

Mujika I, Santisteban J, Castagna C. In-season effect of short-term sprint and power training programs on elite junior soccer players. J Strength Cond Res. 2009;23(9):2581–7.

Thomas K, French D, Hayes PR. The effect of two plyometric training techniques on muscular power and agility in youth soccer players. J Strength Cond Res. 2009;23(1):332–5.

Aagaard P, Andersen JL. Effects of strength training on endurance capacity in top-level endurance athletes. Scand J Med Sci Sports. 2010;20:39–47.

Lopez-Segovia M, Palao Andres JM, Gonzalez-Badillo JJ. Effect of 4 months of training on aerobic power, strength, and acceleration in two under-19 soccer teams. J Strength Cond Res. 2010;24(10):2705–14.

Maio Alves JM, Rebelo AN, Abrantes C, Sampaio J. Short-term effects of complex and contrast training in soccer players’ vertical jump, sprint, and agility abilities. J Strength Cond Res. 2010;24(4):936–41.

Wong PL, Chaouachi A, Chamari K, Dellal A, Wisloff U. Effect of preseason concurrent muscular strength and high-intensity interval training in professional soccer players. J Strength Cond Res. 2010;24(3):653–60.

Sedano S, Matheu A, Redondo JC, Cuadrado G. Effects of plyometric training on explosive strength, acceleration capacity and kicking speed in young elite soccer players. J Sports Med Phys Fitness. 2011;51(1):50–8.

CAS   PubMed   Google Scholar  

Loturco I, Ugrinowitsch C, Tricoli V, Pivetti B, Roschel H. Different loading schemes in power training during the preseason promote similar performance improvements in Brazilian elite soccer players. J Strength Cond Res. 2013;27(7):1791–7.

Los Arcos A, Yanci J, Mendiguchia J, Salinero JJ, Brughelli M, Castagna C. Short-term training effects of vertically and horizontally oriented exercises on neuromuscular performance in professional soccer players. Int J Sports Physiol Perform. 2014;9(3):480–8.

Jovanovic M, Sporis G, Omrcen D, Fiorentini F. Effects of speed, agility, quickness training method on power performance in elite soccer players. J Strength Cond Res. 2011;25(5):1285–92.

Ebben WP, Watts P. A review of combined weight training and plyometric training modes: Complex training. Strength Cond J. 1998;20:18–27.

Article   Google Scholar  

Saez de Villarreal E, Requena B, Izquierdo M, Gonzalez-Badillo JJ. Enhancing sprint and strength performance: combined versus maximal power, traditional heavy-resistance and plyometric training. J Sci Med. 2013;16(2):146–50.

Harris N, Cronin J, Keogh J. Contraction force specificity and its relationship to functional performance. J Sports Sci. 2007;25(2):201–12.

Billot M, Martin A, Paizis C, Cometti C, Babault N. Effects of an electrostimulation training program on strength, jumping, and kicking capacities in soccer players. J Strength Cond Res. 2010;24(5):1407–13.

Young W. Transfer of strength and power training to sports performance. Int J Sports Physiol Perform. 2006;1:74–83.

Schmidtbleicher D. Training for power events. In: Chem PV, editor. Strength and Power in Sports. Boston: Blackwell Scientific; 1992. p. 381-95

Cormie P, McGuigan MR, Newton RU. Developing maximal neuromuscular power: Part 1–biological basis of maximal power production. Sports Med. 2011;41(1):17–38.

Gabriel DA, Kamen G, Frost G. Neural adaptations to resistive exercise: mechanisms and recommendations for training practices. Sports Med. 2006;36(2):133–49.

Toigo M, Boutellier U. New fundamental resistance exercise determinants of molecular and cellular muscle adaptations. Eur J Appl Physiol. 2006;97(6):643–63.

Hakkinen K, Alen M, Kraemer WJ, Gorostiaga E, Izquierdo M, Rusko H, et al. Neuromuscular adaptations during concurrent strength and endurance training versus strength training. Eur J Appl Physiol. 2003;89(1):42–52.

Hakkinen K, Komi PV, Alen M. Effect of explosive type strength training on isometric force- and relaxation-time, electromyographic and muscle fibre characteristics of leg extensor muscles. Acta Physiol Scand. 1985;125(4):587–600.

Aagaard P, Simonsen EB, Andersen JL, Magnusson P, Dyhre-Poulsen P. Increased rate of force development and neural drive of human skeletal muscle following resistance training. J Appl Physiol. 2002;93(4):1318–26.

Helgerud J, Rodas G, Kemi OJ, Hoff J. Strength and endurance in elite football players. Int J Sports Med. 2011;32(9):677–82.

McGawley K, Andersson PI. The order of concurrent training does not affect soccer-related performance adaptations. Int J Sports Med. 2013;34(11):983–90.

Requena B, Gonzalez-Badillo JJ, de Villareal ES, Ereline J, Garcia I, Gapeyeva H, et al. Functional performance, maximal strength, and power characteristics in isometric and dynamic actions of lower extremities in soccer players. J Strength Cond Res. 2009;23(5):1391–401.

Cometti G, Maffiuletti NA, Pousson M, Chatard JC, Maffulli N. Isokinetic Strength and Anaerobic Power of Elite, Subelite and Amateur Soccer Players. Int J Sports Med. 2001;22:45–51.

Faude O, Koch T, Meyer T. Straight sprinting is the most frequent action in goal situations in professional football. J Sports Sci. 2012;30(7):625–31.

Osgnach C, Poser S, Bernardini R, Rinaldo R, di Prampero PE. Energy cost and metabolic power in elite soccer: a new match analysis approach. Med Sci Sports Exerc. 2010;42(1):170–8.

Gaudino P, Iaia FM, Alberti G, Strudwick AJ, Atkinson G, Gregson W. Monitoring training in elite soccer players: systematic bias between running speed and metabolic power data. Int J Sports Med. 2013;34(11):963–8.

Wisloff U, Castagna C, Helgerud J, Jones R, Hoff J. Strong correlation of maximal squat strength with sprint performance and vertical jump height in elite soccer players. Br J Sports Med. 2004;38(3):285–8.

Article   PubMed Central   CAS   PubMed   Google Scholar  

Keiner M, Sander A, Wirth K, Schmidtbleicher D. Long-term strength training effects on change-of-direction sprint performance. J Strength Cond Res. 2014;28(1):223–31.

Silva JR, Magalhaes JF, Ascensao AA, Oliveira EM, Seabra AF, Rebelo AN. Individual match playing time during the season affects fitness-related parameters of male professional soccer players. J Strength Cond Res. 2011;25(10):2729–39.

Dauty M, Potiron Josse M. Correlations and differences of performance between soccer players, professionals, young players and amateurs, from the 10-meter sprint test and knee isokinetic assessment. Sci Sports. 2004;19:75–9.

Koundourakis NE, Androulakis N, Spyridaki EC, Castanas E, Malliaraki N, Tsatsanis C, et al. Effect of different seasonal strength training protocols on circulating androgen levels and performance parameters in professional soccer players. Hormones. 2014;13(1):578–83.

Sander A, Keiner M, Wirth K, Schmidtbleicher D. Influence of a 2-year strength training programme on power performance in elite youth soccer players. Eur J Sport Sci. 2013;13(5):445–51.

Salaj S, Markovic G. Specificity of jumping, sprinting, and quick change-of-direction motor abilities. J Strength Cond Res. 2011;25(5):1249–55.

Helgerud J, Kemi OJ, Hoff J. Pre-season concurrent strength and endurance development in elite soccer players. In: Hoff J, Helgerud J, editors. Football (soccer): new developments in physical training research. Trondheim: NTNU; 2002. p. 55–66.

Brughelli M, Cronin J, Levin G, Chaouachi A. Understanding change of direction ability in sport: a review of resistance training studies. Sports Med. 2008;38(12):1045–63.

Caldwell BP, Peters DM. Seasonal variation in physiological fitness of a semiprofessional soccer team. J Strength Cond Res. 2009;23(5):1370–7.

Requena B, Saez-Saezde Villarreal E, Gapeyeva H, Ereline J, Garcia I, Paasuke M. Relationship between postactivation potentiation of knee extensor muscles, sprinting and vertical jumping performance in professional soccer players. J Strength Cond Res. 2011;25(2):367–73.

Reilly T, Rigby M. Effect of an warm-down following competitive soccer. In: Spinks W, Reilly T, Murphy J, editors. Science and Football IV. London/New York: E&F. N. Spon; 2002. p. 226–9.

Mohr M, Krustrup P, Bangsbo J. Fatigue in soccer: a brief review. J Sports Sci. 2005;23(6):593–9.

Bangsbo J, Iaia FM, Krustrup P. Metabolic response and fatigue in soccer. Int J Sports Physiol Perform. 2007;2(2):111–27.

Reilly T, Drust B, Clarke N. Muscle fatigue during football match-play. Sports Med. 2008;38(5):357–67.

Rampinini E, Bosio A, Ferraresi I, Petruolo A, Morelli A, Sassi A. Match-related fatigue in soccer players. Med Sci Sports Exerc. 2011;43(11):2161–70.

Magalhaes J, Rebelo A, Oliveira E, Silva JR, Marques F, Ascensao A. Impact of Loughborough Intermittent Shuttle Test versus soccer match on physiological, biochemical and neuromuscular parameters. Eur J Appl Physiol. 2010;108(1):39–48.

Rebelo N, Krustrup P, Soares J, Bangsbo J. Reduction in intermittent exercise performance during a soccer match. J Sport Sci. 1998;16:482–3.

Silva JR, Ascensao A, Marques F, Seabra A, Rebelo A, Magalhaes J. Neuromuscular function, hormonal and redox status and muscle damage of professional soccer players after a high-level competitive match. Eur J Appl Physiol. 2013;113(9):2193–201.

Krustrup P, Ortenblad N, Nielsen J, Nybo L, Gunnarsson TP, Iaia FM, et al. Maximal voluntary contraction force, SR function and glycogen resynthesis during the first 72 h after a high-level competitive soccer game. Eur J Appl Physiol. 2011;111(12):2987–95.

Thompson D, Nicholas CW, Williams C. Muscular soreness following prolonged intermittent high-intensity shuttle running. J Sports Sci. 1999;17(5):387–95.

Rahnama N, Reilly T, Lees A, Graham-Smith P. Muscle fatigue induced by exercise simulating the work rate of competitive soccer. J Sports Sci. 2003;21(11):933–42.

Andersson H, Raastad T, Nilsson J, Paulsen G, Garthe I, Kadi F. Neuromuscular fatigue and recovery in elite female soccer: effects of active recovery. Med Sci Sports Exerc. 2008;40(2):372–80.

Greig M. The influence of soccer-specific fatigue on peak isokinetic torque production of the knee flexors and extensors. Am J Sports Med. 2008;36(7):1403–9.

Ispirlidis I, Fatouros IG, Jamurtas AZ, Nikolaidis MG, Michailidis I, Douroudos I, et al. Time-course of changes in inflammatory and performance responses following a soccer game. Clin J Sports Med J Acad Sport Med. 2008;18(5):423–31.

Gaudino P, Gaudino C, Alberti G, Minetti AE. Biomechanics and predicted energetics of sprinting on sand: hints for soccer training. J Sci Med Sport. 2013;16(3):271–5.

Fyfe JJ, Bishop DJ, Stepto NK. Interference between concurrent resistance and endurance exercise: molecular bases and the role of individual training variables. Sports Med. 2014;44(6):743–62.

Apro W, Wang L, Ponten M, Blomstrand E, Sahlin K. Resistance exercise induced mTORC1 signaling is not impaired by subsequent endurance exercise in human skeletal muscle. Am J Physiol Endocrinol Metab. 2013;305(1):E22–32.

Docherty D, Sporer B. A proposed model for examining the interference phenomenon between concurrent aerobic and strength training. Sports Med. 2000;30(6):385–94.

Kraemer WJ, Patton JF, Gordon SE, Harman EA, Deschenes MR, Reynolds K, et al. Compatibility of high-intensity strength and endurance training on hormonal and skeletal muscle adaptations. J Appl Physiol. 1995;78(3):976–89.

Spiering BA, Kraemer WJ, Anderson JM, Armstrong LE, Nindl BC, Volek JS, et al. Resistance exercise biology: manipulation of resistance exercise programme variables determines the responses of cellular and molecular signalling pathways. Sports Med. 2008;38(7):527–40.

Hawley JA. Molecular responses to strength and endurance training: are they incompatible? Appl Physiol Nutr Metab. 2009;34(3):355–61.

Wang L, Mascher H, Psilander N, Blomstrand E, Sahlin K. Resistance exercise enhances the molecular signaling of mitochondrial biogenesis induced by endurance exercise in human skeletal muscle. J Appl Physiol. 2011;111(5):1335–44.

Lundberg TR, Fernandez-Gonzalo R, Gustafsson T, Tesch PA. Aerobic exercise alters skeletal muscle molecular responses to resistance exercise. Med Sci Sports Exerc. 2012;44(9):1680–8.

Lundberg TR, Fernandez-Gonzalo R, Gustafsson T, Tesch PA. Aerobic exercise does not compromise muscle hypertrophy response to short-term resistance training. J Appl Physiol. 2013;114(1):81–9.

Edwards AM, Clark NA. Thermoregulatory observations in soccer match play: professional and recreational level applications using an intestinal pill system to measure core temperature. Br J Sports Med. 2006;40(2):133–8.

Cazzola R, Russo-Volpe S, Cervato G, Cestaro B. Biochemical assessments of oxidative stress, erythrocyte membrane fluidity and antioxidant status in professional soccer players and sedentary controls. Eur J Clin Invest. 2003;33(10):924–30.

Brites FD, Evelson PA, Christiansen MG, Nicol MF, Basilico MJ, Wikinski RW, et al. Soccer players under regular training show oxidative stress but an improved plasma antioxidant status. Clin Sci (Lond). 1999;96(4):381–5.

Article   CAS   Google Scholar  

Egorova ES, Borisova AV, Mustafina LJ, Arkhipova AA, Gabbasov RT, Druzhevskaya AM, et al. The polygenic profile of Russian football players. J Sports Sci. 2014;32(13):1286–93.

Pimenta EM, Coelho DB, Veneroso CE, Barros Coelho EJ, Cruz IR, Morandi RF, et al. Effect of ACTN3 Gene on Strength and Endurance in Soccer Players. J Strength Cond Res. 2013;27(12):3286–92.

Mustafina LJ, Naumov VA, Cieszczyk P, Popov DV, Lyubaeva EV, Kostryukova ES, et al. AGTR2 gene polymorphism is associated with muscle fibre composition, athletic status and aerobic performance. Exp Physiol. 2014;99(8):1042–52.

Elliott MC, Wagner PP, Chiu L. Power athletes and distance training: physiological and biomechanical rationale for change. Sports Med. 2007;37(1):47–57.

Reid MB. Invited Review: redox modulation of skeletal muscle contraction: what we know and what we don’t. J Appl Physiol. 2001;90(2):724–31.

McNamara JM, Stearne DJ. Effect of concurrent training, flexible nonlinear periodization, and maximal-effort cycling on strength and power. J Strength Cond Res. 2013;27(6):1463–70.

Meckel Y, Nemet D, Bar-Sela S, Radom-Aizik S, Cooper DM, Sagiv M, et al. Hormonal and inflammatory responses to different types of sprint interval training. J Strength Cond Res. 2011;25(8):2161–9.

Wahl P. Hormonal and Metabolic Responses to High Intensity Interval Training. J Sports Med Doping Stud. 2013;3(1):e132.

Wahl P, Mathes S, Achtzehn S, Bloch W, Mester J. Active vs. passive recovery during high-intensity training influences hormonal response. Int J Sports Med. 2014;35(7):583–9.

Wahl P, Mathes S, Kohler K, Achtzehn S, Bloch W, Mester J. Acute metabolic, hormonal, and psychological responses to different endurance training protocols. Horm Metab Res. 2013;45(11):827–33.

Zinner C, Wahl P, Achtzehn S, Reed JL, Mester J. Acute hormonal responses before and after 2 weeks of HIT in well trained junior triathletes. Int J Sports Med. 2014;35(4):316–22.

Gunnarsson TP, Christensen PM, Holse K, Christiansen D, Bangsbo J. Effect of additional speed endurance training on performance and muscle adaptations. Med Sci Sports Exerc. 2012;44(10):1942–8.

Ingebrigtsen J, Shalfawi SA, Tonnessen E, Krustrup P, Holtermann A. Performance effects of 6 weeks of aerobic production training in junior elite soccer players. J Strength Cond Res. 2013;27(7):1861–7.

Iaia FM, Rampinini E, Bangsbo J. High-intensity training in football. Int J Sports Physiol Perform. 2009;4(3):291–306.

Wahl P, Guldner M, Mester J. Effects and sustainability of a 13-day high-intensity shock microcycle in soccer. J Sports Sci Med. 2014;13(2):259–65.

PubMed Central   PubMed   Google Scholar  

Gibala MJ, Little JP, van Essen M, Wilkin GP, Burgomaster KA, Safdar A, et al. Short-term sprint interval versus traditional endurance training: similar initial adaptations in human skeletal muscle and exercise performance. J Physiol. 2006;575(Pt 3):901–11.

Burgomaster KA, Hughes SC, Heigenhauser GJ, Bradwell SN, Gibala MJ. Six sessions of sprint interval training increases muscle oxidative potential and cycle endurance capacity in humans. J Appl Physiol. 2005;98(6):1985–90.

Gibala M. Molecular responses to high-intensity interval exercise. Appl Physiol Nutr Metab. 2009;34(3):428–32.

Jensen L, Bangsbo J, Hellsten Y. Effect of high intensity training on capillarization and presence of angiogenic factors in human skeletal muscle. J Physiol. 2004;557(Pt 2):571–82.

Christensen PM, Krustrup P, Gunnarsson TP, Kiilerich K, Nybo L, Bangsbo J. VO2 kinetics and performance in soccer players after intense training and inactivity. Med Sci Sports Exerc. 2011;43(9):1716–24.

Dupont G, Akakpo K, Berthoin S. The effect of in-season, high-intensity interval training in soccer players. J Strength Cond Res. 2004;18(3):584–9.

Cicioni-Kolsky D, Lorenzen C, Williams MD, Kemp JG. Endurance and sprint benefits of high-intensity and supramaximal interval training. Eur J Sport Sci. 2013;13(3):304–11.

Issurin VB. Training transfer: scientific background and insights for practical application. Sports Med. 2013;43(8):675–94.

Reilly T. Energetics of high-intensity exercise (soccer) with particular reference to fatigue. J Sports Sci. 1997;15(3):257–63.

Bangsbo J, Mohr M, Krustrup P. Physical and metabolic demands of training and match-play in the elite football player. J Sports Sci. 2006;24(7):665–74.

Stolen T, Chamari K, Castagna C, Wisloff U. Physiology of soccer: an update. Sports Med. 2005;35(6):501–36.

Kraemer WJ, French DN, Paxton NJ, Hakkinen K, Volek JS, Sebastianelli WJ, et al. Changes in exercise performance and hormonal concentrations over a big ten soccer season in starters and nonstarters. J Strength Cond Res. 2004;18(1):121–8.

Akenhead R, Hayes PR, Thompson KG, French D. Diminutions of acceleration and deceleration output during professional football match play. J Sci Med Sport. 2013;16(6):556–61.

Nummela AT, Paavolainen LM, Sharwood KA, Lambert MI, Noakes TD, Rusko HK. Neuromuscular factors determining 5 km running performance and running economy in well-trained athletes. Eur J Appl Physiol. 2006;97(1):1–8.

Saunders PU, Telford RD, Pyne DB, Peltola EM, Cunningham RB, Gore CJ, et al. Short-term plyometric training improves running economy in highly trained middle and long distance runners. J Strength Cond Res. 2006;20(4):947–54.

Paavolainen L, Hakkinen K, Hamalainen I, Nummela A, Rusko H. Explosive-strength training improves 5-km running time by improving running economy and muscle power. J Appl Physiol. 1999;86(5):1527–33.

Paavolainen LM, Nummela AT, Rusko HK. Neuromuscular characteristics and muscle power as determinants of 5-km running performance. Med Sci Sports Exerc. 1999;31(1):124–30.

Castagna C, Impellizzeri F, Chamari K, Carlomagno D, Rampinini E. Aerobic fitness and Yo-Yo continous and intermittent tests performances in soccer players: A correlation study. J Strength Cond Res. 2006;20(2):320–5.

Brocherie F, Girard O, Forchino F, Al Haddad H, Dos Santos GA, Millet GP. Relationships between anthropometric measures and athletic performance, with special reference to repeated-sprint ability, in the Qatar national soccer team. J Sports Sci. 2014;17:1–12.

Arnason A, Sigurdsson SB, Gudmundsson A, Holme I, Engebretsen L, Bahr R. Physical fitness, injuries, and team performance in soccer. Med Sci Sports Exerc. 2004;36(2):278–85.

Sporis G, Jovanovic M, Omrcen D, Matkovic B. Can the official soccer game be considered the most important contribution to player’s physical fitness level? J Sports Med Phys Fitness. 2011;51(3):374–80.

Silva JR, Magalhaes J, Ascensao A, Seabra AF, Rebelo AN. Training status and match activity of professional soccer players throughout a season. J Strength Cond Res. 2013;27(1):20–30.

Meister S, Faude O, Ammann T, Schnittker R, Meyer T. Indicators for high physical strain and overload in elite football players. Scand J Med Sci Sports. 2013;23(2):156–63.

Tonnessen E, Hem E, Leirstein S, Haugen T, Seiler S. Maximal aerobic power characteristics of male professional soccer players, 1989-2012. Int J Sports Physiol Perform. 2013;8(3):323–9.

Haugen TA, Tonnessen E, Seiler S. Anaerobic performance testing of professional soccer players 1995-2010. Int J Sports Physiol Perform. 2013;8(2):148–56.

Hill-Haas SV, Dawson B, Impellizzeri FM, Coutts AJ. Physiology of small-sided games training in football: a systematic review. Sports Med. 2011;41(3):199–220.

Owen AL, Wong del P, McKenna M, Dellal A. Heart rate responses and technical comparison between small- vs. large-sided games in elite professional soccer. J Strength Cond Res. 2011;25(8):2104–10.

Reinke S, Karhausen T, Doehner W, Taylor W, Hottenrott K, Duda GN, et al. The influence of recovery and training phases on body composition, peripheral vascular function and immune system of professional soccer players. PLoS One. 2009;4(3):e4910.

Article   PubMed Central   PubMed   CAS   Google Scholar  

Sotiropoulos A, Travlos AK, Gissis I, Souglis AG, Grezios A. The effect of a 4-week training regimen on body fat and aerobic capacity of professional soccer players during the transition period. J Strength Cond Res. 2009;23(6):1697–703.

Koundourakis NE, Androulakis NE, Malliaraki N, Tsatsanis C, Venihaki M, Margioris AN. Discrepancy between Exercise Performance, Body Composition, and Sex Steroid Response after a Six-Week Detraining Period in Professional Soccer Players. PLoS One. 2014;9(2):e87803.

Mohr M, Krustrup P, Bangsbo J. Physiological characteristics and exaustive exercise performance of elite soccer players. Med Sci Sports Exerc. 2002;34(5):S24. Supplement 1, May.

Krustrup P, Mohr M, Nybo L, Jensen JM, Nielsen JJ, Bangsbo J. The Yo-Yo IR2 test: physiological response, reliability, and application to elite soccer. Med Sci Sports Exerc. 2006;38(9):1666–73.

Ostojic S. Seasonal alterations in body composition and sprint performance of elite soccer players. J Exerc Physiol Online. 2003;6(3):24–7.

Filaire E, Lac G, Pequignot JM. Biological, hormonal, and psychological parameters in professional soccer players throughout a competitive season. Percept Mot Skills. 2003;97(3 Pt 2):1061–72.

Kraemer WJ, Ratamess NA. Hormonal responses and adaptations to resistance exercise and training. Sports Med. 2005;35(4):339–61.

Schmikli SL, de Vries WR, Brink MS, Backx FJ. Monitoring performance, pituitary-adrenal hormones and mood profiles: how to diagnose non-functional over-reaching in male elite junior soccer players. Br J Sports Med. 2012;46(14):1019–23.

Komi PV, Gollhofer A. Stretch reflexes can have an important role in force enhancement during SSC exercise. J Appl Biomech. 1997;13:451–60.

Bangsbo J. The physiology of soccer - with special reference to intense intermittent exrcise. Acta Physiol Scand. 1994;150:615. suppl.

Wilson JM, Flanagan EP. The role of elastic energy in activities with high force and power requirements: a brief review. J Strength Cond Res. 2008;22(5):1705–15.

Download references

Acknowledgments

No funding was used to assist in the preparation of this review.

Author information

Authors and affiliations.

National Sports Medicine Programme Excellence in Football Project, Aspetar-Qatar Orthopaedic and Sports Medicine Hospital, P.O BOX 29222, Doha, Qatar

João R Silva & George P Nassis

Center of Research, Education, Innovation and Intervention in Sport (CIFI2D), Porto, Portugal

João R Silva & Antonio Rebelo

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to João R Silva .

Additional information

Competing interests.

The authors declare that they have no competing interests.

Authors' contributions

All authors read and approved the final manuscript.

Additional file

Additional file 1: figure s1..

The gains in strength and sprint performance of high-level players after 5 to 10 weeks. Squares represent the 10-m distance [ 2 , 22 , 37 , 38 ]; circles represent the 20-m distance [ 37 ]; rhombi represent the 40-m distance [ 1 , 2 , 6 ]; + symbols represent the average of all distances; triangles represent the average of the 10-m distance; and lines represent the average of the 40-m distance. Figure S2. The gains in strength and jump performance of high-level players after 6 to 10 weeks. Squares represent the squat jump performance (SJ) [ 1 , 6 , 14 , 22 ]; triangles represent the countermovement jump (CMJ) performance [ 2 , 22 , 37 ]; rhombi represent the four bounce test (4BT) performance [ 6 ]; lines represent the five jump test [ 14 ]; circles represent the average CMJ; x symbols represent the average SJ performance; and + symbols represent the average 4BT performance. Figure S3. The gains in strength and change of direction ability of high-level players after 5 to 6 weeks. Squares represent the t -test performance [ 2 , 38 ]; circles represent the Zig-Zag test performance [ 2 ]; and rhombi represent the Illinois agility test performance [ 2 ]. Red-filled triangles represent average of all tests. Figure S4. The gains in strength and overall sprint performance of high-level players following traditional resistance exercise programs (TRE; 6 to 10 weeks) and combined programs (COM; 5 to 7 weeks). Filled circles represent the TRE results; empty circles represent the COM results; red-filled circles represent the average TRE [ 1 , 2 , 37 ]; empty red circles represent the average COM [ 6 , 22 , 38 ]. Figure S5. The gains in strength and overall jump ability of high-level players following traditional resistance exercise programs (TRE; 6 to 10 weeks) and combined programs (COM; 6 to 7 weeks). Blue-filled and unfilled triangles represent the countermovement jump (CMJ) results after TRE and COM, respectively; red-filled and unfilled triangles represent the squat jump (SJ) results after TRE and COM, respectively; green-filled and unfilled triangles represent the four bounce test (4BT) results after TRE and COM, respectively; yellow-filled triangles represent the five jump test (5JT) results after TRE; blue-filled and unfilled circles represent the average CMJ results after TRE [ 2 , 37 ] and COM [ 22 ], respectively; red-filled and unfilled circles represent the average SJ results after TRE [ 1 , 14 ] and COM [ 6 , 22 ], respectively; black-filled and unfilled circles represent the average overall jump ability increases after TRE [ 1 , 2 , 6 , 14 , 37 ] and COM [ 6 , 22 ], respectively.

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License ( https://creativecommons.org/licenses/by/4.0 ), which permits use, duplication, adaptation, distribution, and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Reprints and permissions

About this article

Cite this article.

Silva, J.R., Nassis, G.P. & Rebelo, A. Strength training in soccer with a specific focus on highly trained players. Sports Med - Open 1 , 17 (2015). https://doi.org/10.1186/s40798-015-0006-z

Download citation

Received : 23 July 2014

Accepted : 02 December 2014

Published : 02 April 2015

DOI : https://doi.org/10.1186/s40798-015-0006-z

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Strength Training
• Endurance Training
• Soccer Player
• Sprint Performance