Journal of Sports Science and Medicine
Journal of Sports Science and Medicine
ISSN: 1303 - 2968   
Ios-APP Journal of Sports Science and Medicine
Androit-APP Journal of Sports Science and Medicine
Views
2356
Download
1149
 
©Journal of Sports Science and Medicine (2023) 22, 707 - 725   DOI: https://doi.org/10.52082/jssm.2023.707

Research article
The Success Factors of Rest Defense in Soccer – A Mixed-Methods Approach of Expert Interviews, Tracking Data, and Machine Learning
Leander Forcher1,2, , Leon Forcher1,2, Stefan Altmann1,3, Darko Jekauc1, Matthias Kempe4  
Author Information
1 Institute of Sports and Sports Science, Karlsruhe Institute of Technology, Karlsruhe, Germany
2 TSG 1899 Hoffenheim, Zuzenhausen, Germany
3 TSG ResearchLab gGmbH, Zuzenhausen, Germany
4 Center for Human Movement Sciences, University of Groningen, University Medical Center Groningen (UMCG), Groningen, The Netherlands

Leander Forcher
✉ Institute of Sports and Sports Science, Engler-Bunte-Ring 15, 76133 Karlsruhe, Karlsruhe Institute of Technology, Karlsruhe, Germany
Email: leander.forcher@kit.edu
Publish Date
Received: 15-06-2023
Accepted: 27-10-2023
Published (online): 01-12-2023
 
ABSTRACT

While the tactical behavior of soccer players differs between specific phases of play (offense, defense, offensive transition, defensive transition), little is known about successful behavior of players during defensive transition (switching behavior from offense to defense). Therefore, this study aims to analyze the group tactic of rest defense (despite in ball possession, certain players safeguard quick counterattacks in case of ball loss) in defensive transition. A mixed-methods approach was used, involving both qualitative and quantitative analysis. Semi-structured expert interviews with seven professional soccer coaches were conducted to define rest defense. In the quantitative analysis, several KPIs were calculated, based on tracking and event data of 153 games of the 2020/21 German Bundesliga season, to predict the success of rest defense situations in a machine learning approach. The qualitative interviews indicated that rest defense can be defined as the positioning of the deepest defenders during ball possession to prevent an opposing counterattack after a ball loss. For instance, the rest defending players created a numerical superiority of 1.69 ± 1.00 and allowed a space control of the attacking team of 11.51 ± 9.82 [%] in the area of rest defense. The final machine learning model showed satisfactory prediction performance of the success of rest defense (Accuracy: 0.97, Precision: 0.73, f1-Score: 0.64, AUC: 0.60). Analysis of the individual KPIs revealed insights into successful behavior of players in rest defense, including controlling deep spaces and dangerous counterattackers. The study concludes regaining possession as fast as possible after a ball loss is the most important success factor in defensive transition.

Key words: Team sports, performance analysis, tactics, defensive play, football


           Key Points
  • Combination of qualitative expert interviews and up-to-date quantitative data analysis using tracking data and machine learning revealed insightful results of successful tactical behavior in defensive transition in soccer
  • According to experts, rest defense can be defined as behavior of the deepest defending players during ball possession with the goal to prevent an opposing counterattack after a ball loss during defensive transition
  • To be successful in defensive transition, players in rest defense should control deep spaces and dangerous counterattackers to successfully prevent dangerous opposing counterattacks
  • Most important success criterion in defensive transition is to regain possession after a ball loss as quickly as possible to stop an opponent’s counterattack in the early stages

INTRODUCTION

In soccer practice, it is well known that a single attack can make the difference between victory and defeat. It is quite common for one team to seemingly dominate the other, failing to take the lead and leaving the other team waiting for a decisive counterattack. To prevent this from happening, most teams station several players behind the ball to safeguard a possible counterattack in the event of a sudden ball loss. This tactical approach is referred to as rest defense. While it is widely acknowledged in soccer practice, there is only scarce research on it. To close this gap, this study aims to provide a definition of rest defense that enables a quantification of success, and to identify variables that distinguish between unsuccessful and successful tactical behavior of rest defense.

While soccer performance can be analyzed in the sub-areas of physical, technical, psychological, and tactical performance (Hohmann, 1983; Weineck, 2007), this study will assess the tactical behavior of soccer players in the outlined match situation.

In particular, the tactical behavior of the players differs depending on the playing phase of the game and their respective goals. Accordingly, the match situations in a game can be divided into four distinct playing phases (offensive play, defensive play, defensive transition, offensive transition) which occur in sequence depending on the course of the game (Hewitt et al., 2016; Navarro, 2018). In offensive play, a team controls the ball with the goal of attacking the opponent’s goal to eventually score. In defensive play, a team organizes itself to prevent the opposing team in ball possession from scoring and, in the best case, to regain the ball. The transition phases describe the switch between offensive and defensive playing phases after losing the ball (defensive transition) or gaining the ball (offensive transition). Consequently, there are differences in the behavior of players depending on the playing phase of the game, which should be taken into account when analyzing tactical performance.

Compared to the other playing phases, offensive play has received the most attention from practitioners and researchers. In detail, studies have predominantly investigated offensive on-the-ball actions such as shots (Lucey et al., 2015; Rathke, 2017; Anzer and Bauer, 2021) or passes (Szczepański and McHale, 2016; Power et al., 2017; Chawla et al., 2017). However, tracking data is particularly valuable for analyzing defensive play, as players' behavior off the ball is recorded and can thus be analyzed. Furthermore, defensive play has been revealed to be at least as important for a team's success as offensive play (Georgievski et al., 2019; Lepschy et al., 2021). Consequently, the number of studies analyzing defensive play has increased (Forcher et al., 2022a). For instance, defensive pressure has been assessed to analyze defensive play (Bojinov and Bornn, 2016; Merckx et al., 2021; Forcher et al., 2022b).

The aforementioned transition playing phases have been shown to be highly important for success in soccer as well. Several studies have indicated that short transitions after a ball gain/ball loss have the highest probability of scoring/conceding a goal (Tenga et al., 2010; Lago-Ballesteros, Lago-Peñas and Rey, 2012; Gonzalez-Rodenas et al., 2015). One reason for these effects is the increased vulnerability of an unbalanced defense after losing the ball (Tenga et al., 2010; Gonzalez-Rodenas et al., 2015). In contrast, research on the defensive transition playing phase is still scarce and will therefore be the focus of this study.

Defensive transition describes the behavior of players after losing the ball to restore the defensive organization and thus reach the defensive phase of the game (Bauer and Anzer, 2021). The main goal of this playing phase is to prevent an opponent’s counterattack by securing dangerous pitch areas and, if possible, directly regain the ball by increasing the defensive pressure in ball proximity (DFB-Akademie, 2022). While the defensive behavior of high pressure in close ball proximity after a ball loss is known as counter-pressing (Navarro, 2018; Bauer and Anzer, 2021), the behavior of players behind the ball protecting dangerous pitch areas (despite being the team in ball possession) is often referred to as rest defense (DFB-Akademie, 2022).

As stated earlier, there has been little research on defensive transition in soccer. Nevertheless, two studies examined defensive transitions in general (Vogelbein, Nopp and Hökelmann, 2014; Casal et al., 2016), and one study investigated the specific group tactical behavior of counter-pressing (Bauer and Anzer, 2021). Vogelbein et al. (2014) analyzed the time it took for a defending team to recover a ball (so-called defensive reaction time) in 306 German Bundesliga matches. They found that top teams recovered the ball faster in defensive transition compared to other teams. Similarly, Casal et al. (2016) showed that the duration of the defensive transition is a valuable key performance indicator for defensive transition. In detail, they predicted the success of the defensive transitions of eight matches of FIFA World Cup 2010 using a notational approach (Casal et al., 2016). Concluding, while these studies provide some initial approaches to analyze defensive transitions, their methodological approach does not allow for a detailed analysis of the tactical behavior of defending players (e.g. specific pressing behavior of defending players).

In contrast, approaching defensive transition in more detail, Bauer and Anzer (2021) automatically identified counter-pressing situations and identified crucial variables for the effectiveness of defensive transitions by analyzing over 4000 matches of the German Bundesliga. For instance, their results revealed that having four or more players behind the ball is important to the success of a defensive transition. This study is a great example of the use of player tracking data to get insights into the tactical behavior of players during the defensive transition phase and on the specific group-level counter-pressing in areas close to the ball.

Overall, the detailed knowledge of player behavior in defensive transition is still limited. Especially, none of the abovementioned studies analyzed the group tactical behavior of the players that are not in ball proximity with the task to protect their own goal by controlling dangerous areas and deny or delay counterattacks. As mentioned above, this tactical principle can be referred to as rest defense in practice. However, there is no scientific and generally accepted definition of rest defense in soccer.

Overall, the tactical behavior of soccer players can be analyzed at different levels according to the number of players involved, ranging from individual (1 player), over group (minimum of 2 players), to team level (11 players) analyses. While several studies have indicated that group level tactical behavior seems to be more insightful compared to team level analyses (Goes, 2023), few is known about the tactical behavior of subgroups of players in the defending playing phases (Forcher et al., 2022a).

Accordingly, the aims of this study are (i) to define the tactical behavior referred to as rest defense and (ii) to identify crucial variables (tactical measures) that are important for the success of the defensive transition phase regarding rest defense.

To achieve this goal, we will use a mixed-methods approach via interviews and an observational study. Using expert interviews, we aim to establish a clear definition of the term rest defense and the variables that characterize it. In a second step, we will use this information to analyze data of professional soccer matches to identify the determinants of successful rest defense using state-of-the-art analytics (tracking data and machine learning).

METHODS

Expert interview

This expert interview was conducted according to the guidelines of the Declaration of Helsinki and approved by the local ethics committee (Karlsruhe Institute of Technology, Germany, 21. January 2022).

Participants

Seven professional soccer coaches were interviewed as experts. They were included according to the following inclusion criteria. To be included in this study, experts had to:

  1. have at least four years of practical experience in a professional soccer club (first division or national team) as a head coach, assistant coach, or match analyst,
  2. hold a UEFA soccer coaching license (at least a UEFA B license),
  3. and work as a coach at the time of the interview or have worked as a coach within the last year.

On average, the considered experts were 36.48 (± 4.48) years old and had 10.68 (± 3.44) years of coaching experience in a professional soccer club. Three experts held a UEFA B-license, three experts held a UEFA A-license, and one expert held a UEFA Pro license.

For the recruitment of the experts, the authors' contacts to German Bundesliga clubs and the German Football Association (DFB) were used to contact the experts directly. Of the eight contacted experts, one dropped out.

Theoretical sampling was used. Accordingly, the sample size was determined based on the knowledge gained by the inclusion of additional participants (Blöbaum, Nölleke and Scheu, 2014). Therefore, we stopped expanding the sample when we could no longer expect to gain additional information.

Procedures

To gain information about rest defense, we used a semi-structured expert interview. The guideline for the interview was developed by the main author (LeaF) according to Blöbaum et al. (2014) and Helfferich (2019). The guideline was then discussed with the four co-authors (LeoF, SA, MK, DJ) and adapted if deemed necessary. Finally, the guideline consisted of:

  1. basic personal information request before the start of the interview,
  2. one icebreaker question at the start of the interview,
  3. three entry-level questions about rest defense in general,
  4. and eight questions about detailed information on rest defense (e.g. goals, characteristics, success) (please see Appendix 1 ).

Prior to the interview, each participant was provided with a privacy policy document and participant information regarding the procedures of this study and was then asked to provide informed consent. The interviews were conducted in individual one-to-one conversations and were conducted online via video call or in person. The interviewer remained the same for each expert interview, which lasted on average 15.45 (± 1.56) minutes.

Analysis

The interviews were audio recorded. Afterward, since the goal of expert interviews was to collect objective information, the interviews were transcribed using the following procedures (Blöbaum et al., 2014):

  1. Conversation pauses, body language or other non-verbal signals were not documented,
  2. the interview was grammatically corrected (e.g. dialect),
  3. and sentences were summarized according to the meaning structure of the statement (e.g. repetitions in direct succession are not reported) and put into grammatically correct order.

A qualitative content analysis according to Mayring was conducted on the basis of the transcripts (Mayring, 2022). This approach integrates quantitative and qualitative analysis methods. First, deductive categories are built according to the topics of the interview (e.g. goals of rest defense). Next, as the largest unit of analysis, the answer to a question was determined. Afterwards, inductive subcategories were formed based on the analysis units and according to the deductive categories. In the end, all analysis steps were summarized in a reduction table including the number of the interview, page, line, question, category (deductive), and subcategory (inductive) (please see Appendix 2). This table was then used to count the frequencies of a certain subcategories in each category (see column quantity in Table 1).

Observational study

This data analysis was conducted according to the guidelines of the Declaration of Helsinki and approved by the local ethics committee (Human and Business Sciences Institute, Saarland University, Germany, identification number: 22-02, 10 January 2022).

Data

An observational study design (post-event) was used for this analysis. Tracking and event data from 153 matches of the 2020/21 German Bundesliga season were analyzed.

The tracking data include the positions of all 22 players on the pitch and the ball. The X- and Y-coordinates are tracked by a semi-automatic multi-camera tracking system (TRACAB, ChyronHego, Melville, NY, USA) with a sampling frequency of 25 Hz. With a high validity (spatial precision of position measurement 0.07-0.18 [cm] RMSE) and a high reliability (between device reliability of total distance covered by players: -0.15 ± 0.37 [%]) this technology has shown to be valid for the analysis of soccer-specific performance (Linke, Link and Lames, 2020).

The event data are annotated based on the official match data catalog of the German Soccer League (DFL) (DFL, 2014) by Sportec Solutions (Sportec Solutions AG, Ismaning, Germany). This catalog defines over 30 events with more than 100 attributes (Bauer and Anzer, 2021).

Data processing

All data processing, visualization, and statistical analysis were performed in Python 3.8 using the NumPy, Pandas, Math, Matplotlib, SciPy, SHAP, and scikit-learn libraries.

First, the tracking and event data were synchronized. Due to inaccuracies in the manual annotation of event data the timestamps and origins of events vary from the tracking data. To effectively combine both types of data, we used a synchronization algorithm that has been shown to provide high accuracy in matching the events of event data with the exact time frame of tracking data (Forcher et al., 2023b).

Second, the tactical formation of a team (e.g. 4-4 - 2) and the individual tactical positions of players (e.g. central defender, wide midfielder) were identified. For the tactical formation, offensive and defensive formations were differentiated depending on the ball possession of the teams (Bialkowski et al., 2016). Further, we defined the tactical formation for three time windows: first half (0-45 min), first interval of second half (45-62.5 min), and second interval of second half (62.5-90 min). Those time windows were used to account for in-game formation changes. It has been shown that 95% of in-game formation changes during the game occurred in the second half (Forcher et al., 2022c). Overall, following this procedure we collected six formations per team per match. We used a formation descriptor based on a KMeans clustering algorithm to cluster the mean longitudinal x-positions of outfield players (excluding the goalkeeper) for the considered time window (e.g. in ball possession and first half) into three formation lines (e.g. 4-3 - 3) (Goes et al., 2021). To define the tactical position, we used the vertical y-positions of players to discriminate between wide and central positions resulting in the following seven possible tactical positions (goalkeeper, central/wide defender, central/wide midfielder, central/wide striker).

Rest defense situations & success

To identify rest defense situations, we used the information from the expert interviews (see section 2 Expert interview). Accordingly, we identified ball possession changes while the ball stayed in play and the ball-gaining team had a minimum of one intentional action on the ball using the event data. This procedure was chosen to exclude unintentional ball possessions by the ball-gaining team (such as ball deflections) where no rest defense game situation occurs. Furthermore, we focused on ball possession changes in the opposing attacking third when the opposing midfielder-line was overplayed (using the mean x-position of the midfielder-line & the x-position of the ball) utilizing the tracking data (see section 2.2 Results: playing phase of rest defense).

To assess rest defense, we considered all players (attackers and defenders) located in the area ten meters in front of the defender closest to the own goal line (see Figure 1).

Furthermore, to define the success of a defensive transition situation we applied the results of the expert interview (please see section 2.2 Results: Successful/unsuccessful rest defense [outcome]). In detail, to focus on the effects of defensive transition, we considered only opposing ball possessions after a considered change of possession with a maximum duration of twelve seconds. This twelve second threshold was determined in accordance with a previous study about defensive transition (Bauer and Anzer, 2021). Accordingly, successful rest defense situations were defined as a ball regain in the following twelve seconds after the identified ball loss. Unsuccessful rest defense situations were defined as an opposing shot on goal in the following twelve seconds after an identified ball loss. All other results of an opposing counterattack (e.g. ball out of play, stoppage of play) were not considered, because according to the expert interview they could not be clearly assigned to either successful or unsuccessful rest defense.

Variables

All variables used to analyze the rest defense situation were measured in the identified moment of a considered ball possession change. This moment of a ball loss was identified as the most important situation of rest defending behavior, as it characterizes the change from the offensive playing phase to the defensive transition. Both of which were indicated in the expert interviews as defining phases of the rest defense (see expert-interview: Playing phase of rest defense).

First, the tactical positions of players in the considered area of rest defense (ten meters in front of the deepest defender, see Figure 1) were considered. This procedure was conducted to investigate whether this approach identified the players with the tactical position, which was also indicated by the experts in the expert interviews (see expert-interview: Players involved in rest defense).

Second, the number of defenders, the number of attackers, and the numerical superiority of the defending team in the defined rest defense area were analysed.

Third, the marking of attackers in this area was analyzed.Thereby, the defensive pressure on attackers (mean, min) was measured from 0-100 [%] using a pressure model of Andrienko et al. (2017) that was expanded by Herold et al. (2022). Further, the closest distance of the defenders to the attackers [m] (mean, max) was determined.

Fourth, the spatial formation of the rest defending players (defenders in the area of rest defense) was measured using the surface area [m2] (area of the convex hull) (Moura et al., 2012), the width and length of the surface area [m], and spread [m] (square root of the sum of squared standard deviation from their average position [centroid]) (Bourbousson et al., 2010; Bartlett et al., 2012) (see Figure 1).

Fifth, the height of the rest defense was computed by calculating the distance of the deepest defender to the team’s own goal-line [m].

Sixth, the space control of both teams from the area of rest defense to the defending team’s goal line was calculated using Voronoi diagrams (absolute [m2] & relative [%], scipy.spatial package in python) (see Figure 1).

Finally, the duration of the rest defending situation (defensive transition), the number of passes, and the number of actions (e.g. tacklings) of the opponent’s counterattack were measured.

Statistical analysis

We deployed several classifiers to best solve our binary problem (successful vs unsuccessful) to predict the success of a rest defense situation (defensive transition). Accordingly, we used logistic regression (ridge regression, elastic net regularization), Random Forest Classifier, Gradient Boosting, XGBoost Classifier, and AdaBoost Classifier and used a train-test-split of 70% and 30%, respectively. To evaluate the performance of the prediction models we calculated the Accuracy, Precision, f1-Score, and Area Under the Curve (AUC). Since our dataset is unbalanced (97 % successful & 3 % unsuccessful), we used f1-Score for model optimization. To gain insights into the dependencies of the prediction we computed Shapley values for the final model.

RESULTS

Expert interview

The results of the expert interviews are presented in Table 1.

According to the experts, the goal of rest defense is to prevent opponent’s counterattacks by securing dangerous areas (e.g. deep areas) or controlling dangerous attackers and to regain the ball. Players proceed into the rest defense when they no longer have tasks in the attack. This happens, for example, in game situations where the ball is controlled in the attacking third or when the defending midfielder-line is overplayed. Accordingly, the players in rest defense position themselves during the attack (i.e. offensive play). However, rest defense only comes into play when the ball is lost and the team is in defensive transition.

Which players are involved in rest defense depends on the tactical formation (e.g. 3-4-3), the opponent (e.g. number of strikers of the opposing team), and the game situation (e.g. ball position). In general, the central defenders, wide defenders (both also referred to as defensive line), and central midfielders are most often involved in rest defense. The most stated tactical approach of rest defense was man-to-man defense. The majority situations (+1 & +2 majority) were also identified, with the differentiation of the tactical approach (sandwich: defenders in front and behind the attacker, flat: all defenders behind the attacker).

Observational study

Overall, we identified n = 2951 rest defense situations. 2425 of them were classified as successful and 75 were classified as unsuccessful which were considered for the prediction model. 451 match situations did not result in a ball regain or a shot on goal and therefore were not considered (e.g. ball out of play, see methods section: Rest defense situations & success).

The tactical playing positions identified in the area of rest defense were 70.1% central defenders (n = 6459), 13.4 % wide defenders (n = 1227), 9.5 % central midfielders (n = 866), 5.9 % wide midfielders (n = 541), and 0.2% wide strikers (n = 19). The identified attackers in the area of rest defense were 78.5 % central strikers (n = 3520), 8.3 % central midfielders (n = 370), 6.5 % wide strikers (n = 291), 5.2 % wide midfielders (n = 234), 1.0 % central defenders (n = 46), and 0.5 % wide defenders (n = 23).

On average, the opponent’s counterattack after the ball loss lasted 10.92 ± 0.61 [s], had 6.41 ± 1.50 actions, and 3.55 ± 1.29 passes.

In the moment of the considered possession change, we identified 3.70 ± 0.76 defenders and 2.01 ± 0.93 attackers in the area of rest defense, resulting in a defensive numerical superiority of 1.69 ± 1.00.

The spatial formation of the rest defending players resulted in a surface area of 85.98 ± 60.09 [m2] which was 7.35 ± 1.92 [m] long and 28.21 ± 7.43 [m] wide. The spread was 13.93 ± 2.86 [m], on average.

On average, the rest defense was positioned 43.56 ± 10.00 [m] ahead of their own goal line.

The mean distance of the closest defender to the attackers was 5.08 ± 2.23 [m] and the longest distance from a defender to the attackers was 6.89 ± 4.19 [m]. This resulted in a mean pressure on the attacking players of 6.66 ± 12.59 [%] and a minimum pressure of 2.77 ± 11.09 [%].

The space control of the attacking team from the area of rest defense to the defending team’s goal line was 390 ± 299.99 [m2] resulting in a ratio of 11.51 ± 9.82 [%] relative to the defensive team.

The results of the classifiers predicting the success of rest defense situations are shown in Table 2. Furthermore, the final model (AdaBoost Classifier [excluding distance variables]) with the best prediction performance (based on f1-Score) was chosen for further analysis. For this model, Shap values were computed which are depicted in Figure 2.

DISCUSSION

The aim of this study was twofold. First, we conducted expert interviews with professional soccer coaches to define the tactical behavior rest defense to establish a working definition that enables further performance analysis. Second, based on the gained knowledge of the interviews, we developed a data analysis to identify critical variables that are important to the success of rest defense.

In employing a mixed-methods research design that integrates both qualitative and quantitative approaches, this study aims to elucidate the conditions that underlie tactical behavior in soccer. Initial qualitative insights were garnered through interviews with subject matter experts, and these data were subsequently employed to inform a nuanced quantitative analysis. The analytical framework was designed to evaluate the success metrics associated to rest defense in the playing phase of defensive transition.

This detailed information gained by the qualitative expert interviews was applied to the quantitative analysis. To illustrate its quality, we provide some general information. The approach to detect rest defense situations identified players with the same tactical positions as those named by the experts as being involved in rest defense, as indicated by a high degree of agreement (93% defensive line & central midfielder). Accordingly, this approach using the area of rest defense seems to be valid for identifying rest defense situations and rest defending players.

The machine learning approach to predict the success of defensive transitions using information about rest defending players showed satisfactory predictive performance. Our final model (AdaBoost [excluding distance variables]) outperformed the previous approach to model the success of defensive transition by Casal et al. (2016), which showed an accuracy of 0.58 (our model: 0.97). In comparison to Bauer and Anzer’s prediction of defensive transition success, our model showed a comparable prediction performance (Precision: 0.72 [our model: 0.73], f1-Score: 0.67 [our model: 0.64]), however, poorer performance in distinguishing between the classes (AUC: 0.87 [our model: 0.60]). Their analysis included information of all players involved in defensive transition, also the behavior of defending players in ball proximity. Recent studies on defensive behavior in soccer showed that especially the behavior of players in close ball proximity is important for defensive success (Forcher et al., 2022b). In contrast, our study used predominantly information about the rest defending players who are not in ball proximity in the moment of a ball loss. Therefore, this study especially indicates the importance of players in rest defense (not neccessarily in ball proximity) for the defensive success in defensive transition.

After demonstrating the quality of our approach, the prediction model is discussed by evaluating each variable according to its value for the prediction. The most important variable in the prediction of rest defense situations during defensive transition was the duration of opposing counterattacks. The analysis of SHAP values (see Figure 2) suggests that a decrease in the duration of counterattacks increases the probability of successful rest defense. This finding is supported by the results of Casal et al. (2016) and Vogelbein et al. (2014), both of which found that the time to recover the ball in defensive transition is an important indicator of successful performance. Furthermore, Bauer and Anzer (Bauer and Anzer, 2021) indicated that the chance of conceding a goal is greatly increased if the ball is not regained within five seconds. Overall, this supports the conclusion that regaining the ball quickly increases the success of the defensive transition by giving the counterattacking team less opportunity to build their attack and deny their actions early.

Moreover, the prediction model yielded further insights that support the conclusion that duration is a critical success factor for defensive transition. In detail, the number of actions and the number of passes of the opposing counterattack were also highly important for the prediction (2nd and 3rd most important variables, see Figure 2). The identified distributions of both variables are comparable to other investigations, which indicates a higher generality of our findings. Specifically, we showed that on average an opposing counterattack that started in the defensive third had between 3-4 passes and 5-6 actions. Comparably, several other studies (of World Cup 2002 & Champion’s League 2014/15) found that successful counterattacks (ending in a scoring opportunity or a goal) required about 4-8 actions (especially when starting from their own defensive third) (Fleig and Hughes, 2004; Hughes and Lovell, 2019). While the variable number of actions showed a similar distribution as the duration of the counterattack (fewer opposing actions increase the probability of success of rest defense), the number of passes showed an opposite trend. This opposing trend could be due to high correlation between the variables pass number, number of actions, and the duration of the counterattacks which could have influenced the results of the prediction model. To sum up, the time it takes a defending team to recover the ball after a ball loss is a crucial success factor in defensive transition.

Besides, the height of the rest defense was the 4th most important variable for the prediction of rest defense success. With it, the distribution of SHAP values in Figure 2 suggests that a deeper rest defense (closer to the own goal line) is beneficial for the success of rest defense. A possible reason for this is the smaller space behind the defense, which decreases the chance for an opposing counterattack to play in behind this deep rest defense. Following this idea, the deep spaces could be better secured by the rest defending players. This is in line with the stated primary goals of rest defense by experts, to safe deep spaces (see Figure 1: Goal of rest defense). In this context, Castellano and Casamichana (2015) showed that better teams in Spanish La Liga positioned their defensive-line closer to their own goal. However, they did not differentiate their analysis according to the playing phase and therefore, comparison to the current results is hardly possible. Moreover, this is an interesting finding when comparing the stated ideas to the group tactic of counter-pressing after a ball loss (in defensive transition). If a team aims to counter-press deep in the opposing attacking third after a ball loss one could argue that it is helpful that the last line is also high up the pitch to decrease the space for the counterattacking team to play and increase the pressure on the opponents. However, our result suggests kind of an opposing trend where it seems to be helpful for the success of defensive transition that the rest defending players are closer to their own goal. This might be a perfect example of risk and reward in a particular game phase in soccer, where opposite tactical behaviors can have both benefit and risk depending on the goal to be achieved (e.g. defensive line moved high up the pitch to enhance counter-pressing, which is assumed to decrease the performance of rest defense). Summing up, the effects of a deeper rest defense enhancing the success in defensive transition should be analyzed in the combination of counter-pressing in ball proximity.

Furthermore, the variable numerical superiority was the 5th most important variable in our prediction with a higher defensive numerical superiority suggesting an increase in the success of rest defense. This finding is supported by the results of Bauer and Anzer (2021) who indicated that it is beneficial for the success in defensive transition when four or more players are behind the ball after a ball loss. This seems intuitive, as more defenders can more easily control possible counterattacking players and dangerous spaces (e.g. deep spaces). Similarly, in the analysis of the playing phase of defensive play, it has been repeatedly shown that numerical superiority is crucial for defensive success (Gréhaigne et al., 2002; Forcher et al., 2023a). On average, we found a defensive majority of + 1.7 (± 1) defenders in the rest defense area, which is comparable to the findings of Vilar et al. (2013) who identified +1 player superiorities in the center-back sub-areas of the defense. While these areas can be interpreted as the rest defense area this study also focused on the playing phase of defensive play. With the present study design, no conclusions can be drawn to the tactical positioning of rest defending players (e.g. flat or sandwich, see expert interview: Tactics of rest defense). However, the variables number of defenders and attackers are less important for the prediction as they are partly already mapped in the superiority measure.

The 6th most important variable for defensive success in rest defense was the ratio of space control of attacking players with respect to the defenders in the area of rest defense to the defending team’s goal line (see Figure 1). In detail, the SHAP values indicate that a low space control of the attacking team enhances defensive success in rest defense. In this regard, numerous studies have investigated measures of space control to analyze tactical match performance in soccer (Filetti et al., 2017; Rein et al., 2017; Fernández et al., 2018; Low et al., 2021). For instance, Rein et al. (2017) found that passes causing a high space control gain in the attacking third are connected to goal scoring. Interpreted conversely, a small space control in front of the own goal is beneficial for defensive success which supports the current findings. By denying space control of the attackers, the defenders control dangerous counterattackers and deep spaces. Therefore, the metric space control appears to optimally quantify the idea of controlling deep spaces and areas around counterattackers. Both principles were stated to be highly important goals for experts in rest defense (see expert interview: Goals of rest defense). However, the results of outliers with extremely high space control of attackers seem to be counterintuitive, suggesting a positive effect on rest defense success (see SHAP values in Figure 2). A possible explanation for this pattern could be counterattackers standing behind the rest defenders (in offside), which would increase the space control tremendously, but decrease the chance of a successful counterattack as they cannot legally intervene in the game situation.

The spatial formation of the rest defense was partially important for the prediction of success in rest defense (7th - 10th most important variables for prediction, see Figure 2). The trend of SHAP values indicates that lower values of spread, surface area, length, and width of rest defending players (higher compactness) are advantageous for success of rest defense. In this context, there have been several studies analyzing the spatial organization (e.g. using surface area or spread) of the whole defending team (Moura et al., 2012; Bartlett et al., 2012; Castellano et al., 2013; Clemente et al., 2013; Welch et al., 2021). While it remains questionable whether higher compactness of the whole defending team is beneficial for defensive success, it was shown that a higher compactness in areas close to the ball could be an important success factor in defense (Forcher et al., 2023b). However, those studies analyzed the playing phase of defensive play and did not focus on subgroups of players that are not in direct ball proximity (such as rest defense) which makes a direct conclusion to the current results difficult.

Finally, defensive pressure showed only a weak predictive performance for rest defense success in comparison to the other metrics in the present study. The identified mean pressure of about 7 [%] is smaller compared to other findings on defensive pressure (11-20 [%] in the playing phase of defensive play (Forcher et al., 2022b), or 8-30 [%] on pass receivers irrespective of the playing phase (Herold et al., 2022)). In this context, it was shown that the defensive pressure decreases in areas further away from the ball (Forcher et al., 2022b). This is especially the case in the current analysis of the tactical behavior of players in the rest defense who are specifically not in ball proximity. In contrast to our findings, defensive pressure was previously shown to be a good indicator for defensive success (Forcher et al., 2022b). This could be due to the other variables (such as space control) that might represent the metric of defensive pressure with other measurement approaches (e.g. high defensive pressure possibly results from large space control). However, higher pressure on the attackers (to deny their actions) in the rest defense area (see Figure 1) seems to increase the probability of a successful transition.

Practical application

The present study provides valuable information that can be practically applied in various circumstances. Specifically, the findings can be utilized in training sessions to focus on the tactical components that have been identified as critical success factors in rest defense. This, in turn, can improve the overall performance of players in this particular group tactic. Coaches are advised to emphasize the positioning of rest defending players to control deep spaces and potential counterattackers by marking them. Additionally, the rest defense should maintain a high level of compactness while allowing defenders to control opposing attackers within the rest defense area. Defensive majority situations may be advantageous as they offer greater control over space around attackers, thus reducing their actions. Ultimately, the defending team should strive to minimize the opponent's actions following ball loss to regain possession as quickly as possible. In conclusion, the principles of play outlined in this study can enhance coaches' understanding of the key aspects of rest defense.

Furthermore, the presented variables used in the quantitative study can be applied to objectively analyze the performance of rest defense during defensive transitions in post-match, live, or opponent analysis. Thereby, the presented prediction model can be applied to evaluate each individual game situation of rest defense by analyzing the identified patterns. Finally, this analysis can help the coaching staff to assess which specific parts of the tactical behavior were beneficial in a specific game situation and what should be adjusted to enhance success of the defensive transition.

Limitations and future research

In addition to the practical implications discussed earlier, the current study has certain limitations. First, our analysis of successful outcomes of defensive transition was limited to ball gains, despite other outcomes, such as the ball going out of play, also being potentially successful. Second, we only examined a specific group tactic (rest defense) in defensive transition, without taking into account the interactions among all eleven players, which can impact the team performance. Future research should focus on combining rest defense with counter-pressing to explore multiple group tactics in defensive transition and their risk and reward trade-offs. Third, our dataset was highly unbalanced, with only 3% of defensive transitions leading to an opposing shot on goal, which may have influenced our results. However, our large dataset and optimization of the machine learning approach using the f1-Score helped mitigate this issue. Fourth, the metric used to quantify pitch control (Voronoi diagrams) was a basic approach that did not account for player orientation and speed. Future studies could explore advanced methods for quantifying space control, given the significance of this factor in defensive transition. In the end, our study solely analyzed the tactical positioning of rest defending players in the crucial moment of ball loss without taking the time intervals before and after the ball loss into account. This is a simplification tactical behavior that should be noted when interpreting the results. However, expert interviews suggested that player behavior in rest defense is essential in both playing phases of offensive play and defensive transition. Therefore, future research could examine the behavior of players in rest defense immediately before and after losing possession of the ball. In the end, the found effects of successful tactical behavior need to be confirmed in future studies (e.g. by detailed video match analysis).

CONCLUSION

This study showed how to combine qualitative and quantitative research in soccer, and how to use expert knowledge to enhance an up-to-date analysis of tactical behavior. With it, we presented practically important knowledge about how to behave in rest defense. Concluding, rest defense is defined as behavior of the deepest defenders during ball possession with the goal to prevent an opposing counterattack after a ball loss during defensive transition. Our results suggest that rest defending players should control deep spaces and dangerous counterattackers to successfully prevent a fast opposing counterattack. This could allow the defensive team to regain possession as quickly as possible in defensive transition to stop an opponent’s counterattack in the early stages, which was shown to be most important for success in defensive transition.

ACKNOWLEDGEMENTS

The authors thank the German Football League (Deutsche Fußball Liga, DFL) for providing the match data used in this study.The authors have no conflict of interest to declare. The present study complies with the current laws of the country in which it was performed. The used data is property of the German Football League (Deutsche Fußball Liga, DFL) and is not publicly available. The authors do not have permission to share the data publicly. This work can be reproduced using similar data from professional soccer (e.g. tracking and event data of other soccer leagues).

AUTHOR BIOGRAPHY

Journal of Sports Science and Medicine Leander Forcher
Employment: Research associate at Karlsruhe Institute of Technology & Match Analyst TSG 1899 Hoffenheim
Degree: M. Sc.
Research interests: Performance analysis, team sports, soccer
E-mail: leander.forcher@kit.edu
 

Journal of Sports Science and Medicine Leon Forcher
Employment: Research associate at Karlsruhe Institute of Technology & Match Analyst at TSG 1899 Hoffenheim
Degree: M.Sc.
Research interests: Performance analysis, team sports, soccer
E-mail: leon.forcher@kit.edu
 

Journal of Sports Science and Medicine Stefan Altmann
Employment: Research associate at Karlsruhe Institute of Technology and TSG ResearchLab, Germany
Degree: Dr.
Research interests: Exercise physiology, performance testing, team sports, soccer
E-mail: stefan.altmann@kit.edu
 

Journal of Sports Science and Medicine Darko Jekauc
Employment: Professor at Karlsruhe Institute of Technology
Degree: Prof. Dr.
Research interests: Health education, sports psychology
E-mail: darko.jekauc@kit.edu
 

Journal of Sports Science and Medicine Matthias Kempe
Employment: Assistant professor at University of Groningen
Degree: Dr.
Research interests: Sports science, sports psychology, computer science
E-mail: m.kempe@umcg.nl
 
 
REFERENCES
Journal of Sports Science and Medicine Andrienko G., Andrienko N., Budziak G., Dykes J., Fuchs G., von Landesberger T., Weber H. (2017) Visual analysis of pressure in football. Data Mining and Knowledge Discovery 31, 1793-1839.  Crossref
Journal of Sports Science and Medicine Anzer G., Bauer P. (2021) A Goal Scoring Probability Model for Shots Based on Synchronized Positional and Event Data in Football (Soccer). Frontiers in Sports and Active Living 3.  Crossref
Journal of Sports Science and Medicine Bartlett R., Button C., Robins M., Dutt-Mazumder A., Kennedy G. (2012) Analysing Team Coordination Patterns from Player Movement Trajectories in Soccer: Methodological Considerations. International Journal of Performance Analysis in Sport 12, 398-424.  Crossref
Journal of Sports Science and Medicine Bauer P., Anzer G. (2021) Data-driven detection of counterpressing in professional football. Data Mining and Knowledge Discovery 35, 2009-2049.  Crossref
Journal of Sports Science and Medicine Bialkowski A., Lucey P., Carr P., Matthews I., Sridharan S., Fookes C. (2016) Discovering Team Structures in Soccer from Spatiotemporal Data. IEEE Transactions on Knowledge and Data Engineering 28, 2596-2605.  Crossref
Journal of Sports Science and Medicine Blöbaum B., Nölleke, D. and Scheu, A.M. (2014) Das Experteninterview in der Kommunikationswissenschaft. In: Handbuch nicht standardisierte Methoden in der Kommunikationswissenschaft. Springer NachschlageWissen. Eds: Averbeck-Lietz, S. and Meyen, M. Wiesbaden: Springer Fachmedien. 1-13.  Crossref
Journal of Sports Science and Medicine Bojinov, I. and Bornn, L. (2016) The pressing game: Optimal defensive disruption in soccer. In: 10th MIT Sloan Sports Analytics Conference. 11.-12.03.2016. Boston, United States of America. 1-8.
Journal of Sports Science and Medicine Bourbousson J., Sève C., McGarry T. (2010) Space-time coordination dynamics in basketball: Part 2. The interaction between the two teams. Journal of Sports Sciences 28, 349-358.  Crossref
Journal of Sports Science and Medicine Casal C.A., Andujar M.Á., Losada J.L., Ardá T., Maneiro R. (2016) Identification of Defensive Performance Factors in the 2010 FIFA World Cup South Africa. Sports (Basel, Switzerland) 4, 54.  Crossref
Journal of Sports Science and Medicine Castellano J., álvarez D., Figueira B., Coutinho D., Sampaio J. (2013) Identifying the effects from the quality of opposition in a Football team positioning strategy. International Journal of Performance Analysis in Sport 13, 822-832.  Crossref
Journal of Sports Science and Medicine Castellano J., Casamichana D. (2015) What are the differences between first and second divisions of Spanish football teams?. International Journal of Performance Analysis in Sport 15, 135-146.  Crossref
Journal of Sports Science and Medicine Chawla S., Estephan J., Gudmundsson J., Horton M. (2017) Classification of Passes in Football Matches Using Spatiotemporal Data. ACM Transactions on Spatial Algorithms and Systems 3, 1-30.  Crossref
Journal of Sports Science and Medicine Clemente F.M., Couceiro M., Martins F., Mendes R., Figueiredo A. (2013) Measuring Collective Behaviour in Football Teams: Inspecting the impact of each half of the match on ball possession. International Journal of Performance Analysis in Sport 13, 678-689.  Crossref
Journal of Sports Science and Medicine DFB-Akademie (2022) Dominanz durch eine effektive Restverteidigung. Available from URL: https://www.dfb-akademie.de/dominanzdurch-eine-effektive-restverteidigung/-/id-11010557 [Accessed 24 February 2023].
Journal of Sports Science and Medicine DFL (2014) Definitionskatalog Offizielle Spieldaten (Definitions for Official Gama Data). DFL: Definitionskatalog Offizielle Spieldaten. Available from URL: https://www.bundesliga.com/de/bundesliga/news/dfl-definitionskatalog-offizielle-spieldaten_0000265687.jsp [Accessed 5 December 2022].
Journal of Sports Science and Medicine Fernández, J., Bornn, L. and Fraser, S. (2018) Wide Open Spaces: A statistical technique for measuring space creation in professional soccer. In: MIT Sloan Sports Analytics Conference. 03- 04.03.2017. Boston, United States of America. 1-19.5687.jsp [Accessed 5 December 2022].
Journal of Sports Science and Medicine Filetti C., Ruscello B., D’Ottavio S., Fanelli V. (2017) A study of relationships among technical, tactical, physical parameters and final outcomes in elite soccer matches as analyzed by a semiautomatic video tracking system. Perceptual and Motor Skills 124, 601-620.  Crossref
Journal of Sports Science and Medicine Fleig, G. and Hughes, M. (2004) Counter attacks in the 2002 World Cup for association football. In: Performance Analysis of Sport VI. Eds: O'Donoghue, P. and Hughes, M. Cardiff: UWIC. 110-122.
Journal of Sports Science and Medicine Forcher L., Altmann S., Forcher L., Jekauc D., Kempe M. (2022a) The use of player tracking data to analyze defensive play in professional soccer - A scoping review. International Journal of Sports Science & Coaching 17, 1567-1592.  Crossref
Journal of Sports Science and Medicine Forcher L., Beckmann T., Wohak O., Romeike C., Graf F., Altmann S. (2023a) Prediction of defensive success in elite soccer using machine learning - Tactical analysis of defensive play using tracking data and explainable AI. Science and Medicine in Football , 1-12.  Crossref
Journal of Sports Science and Medicine Forcher L., Forcher L., Altmann S., Jekauc D., Kempe M. (2022b) The keys of pressing to gain the ball - Characteristics of defensive pressure in elite soccer using tracking data. Science and Medicine in Football , 1-11.  Crossref
Journal of Sports Science and Medicine Forcher L., Forcher L., Altmann S., Jekauc D., Kempe M. (2023b) Is a compact organization important for defensive success in elite soccer? - Analysis based on player tracking data. International Journal of Sports Science & Coaching , 1-12.  Crossref
Journal of Sports Science and Medicine Forcher L., Preine L., Forcher L., Wäsche H., Jekauc D., Woll A., Gross T., Altmann S. (2022c) Shedding some light on in-game formation changes in the German Bundesliga: Frequency, contextual factors, and differences between offensive and defensive formations. International Journal of Sports Science & Coaching , 1-10.  Crossref
Journal of Sports Science and Medicine Georgievski B., Labadze L., Aboelsoud M.E. (2019) Comparative advantage as a success factor in football clubs: Evidence from the English Premier League (EPL). Journal of Human Sport & Exercise 14, 292-314.  Crossref
Journal of Sports Science and Medicine Goes F., Brink M.S., Elferink-Gemser M.T., Kempe M., Lemmink K.A.P.M. (2021) The tactics of successful attacks in professional association football: large-scale spatiotemporal analysis of dynamic subgroups using position tracking data. Journal of Sports Sciences 39, 523-532.  Crossref
Journal of Sports Science and Medicine Goes, F. (2023) Tactical Behaviour in Professional Soccer - The Secret of Successful Attacks. PhD Thesis, UNiversity of Groningen, Groningen.
Journal of Sports Science and Medicine Gonzalez-Rodenas J., Lopez-Bondia I., Calabuig F., Pérez-Turpin J.A., Aranda R. (2015) The effects of playing tactics on creating scoring opportunities in random matches from US Major League Soccer. International Journal of Performance Analysis in Sport 15, 851-872.  Crossref
Journal of Sports Science and Medicine Gréhaigne, J.F., Marchal, D. and Duprat, E. (2002) Regaining possession of the ball in the defensive area in soccer. In: Science and Football IV. Eds: Murphy, A., Reilly, T. and Spinks, W. London, United Kingdom: Routledge
Journal of Sports Science and Medicine Helfferich, C. (2019) Leitfaden- und Experteninterviews. In: Handbuch Methoden der empirischen Sozialforschung. Eds: Baur, N. and Blasius, J. Wiesbaden: Springer Fachmedien. 669-686.  Crossref
Journal of Sports Science and Medicine Herold M., Hecksteden A., Radke D., Goes F., Nopp S., Meyer T., Kempe M. (2022) Off-Ball Behavior in Association Football: A Data-Driven Model to Measure Changes in Defensive Pressure. Journal of Sport Science 40, 1412-1425.  Crossref
Journal of Sports Science and Medicine Hewitt A., Greenham G., Norton K. (2016) Game style in soccer: what is it and can we quantify it?. International Journal of Performance Analysis in Sport 16, 355-372.  Crossref
Journal of Sports Science and Medicine Hohmann A. (1983) Theoretische Aspekte der Leistungsdiagnostik im Sportspiel. Leistungssport 13, 5-10.
Journal of Sports Science and Medicine Hughes M., Lovell T. (2019) Transition to attack in elite soccer. Journal of Human Sport & Exercise 14, 236-253.  Crossref
Journal of Sports Science and Medicine Lago-Ballesteros J., Lago-Peñas C., Rey E. (2012) The effect of playing tactics and situational variables on achieving score-box possessions in a professional soccer team. Journal of Sports Sciences 30, 1455-1461.  Crossref
Journal of Sports Science and Medicine Lepschy H., Woll A., Waesche H. (2021) Success Factors in the FIFA 2018 World Cup in Russia and FIFA 2014 World Cup in Brazil. Frontiers in Psychology 12, 1-9.  Crossref
Journal of Sports Science and Medicine Linke D., Link D., Lames M. (2020) Football-specific validity of TRACAB’s optical video tracking systems. Plos One 15, 1-17.  Crossref
Journal of Sports Science and Medicine Low B., Rein R., Raabe D., Schwab S., Memmert D. (2021) The porous high-press? An experimental approach investigating tactical behaviours from two pressing strategies in football. Journal of Sports Sciences 39, 1-12.  Crossref
Journal of Sports Science and Medicine Lucey P., Bialkowski A., Monfort M., Carr P., Matthews I. (2015) . “Quality vs Quantity”: Improved Shot Prediction in Soccer using Strategic Features from Spatiotemporal Data , -.
Journal of Sports Science and Medicine Mayring, P. (2022) Qualitative Inhaltsanalyse Grundlagen und Techniken. Available from URL: https://rds-blb.ibsbw.de/link?kid=1814287868.
Journal of Sports Science and Medicine Merckx S., Robberechts P., Euvrard Y., Davis J. (2021) Measuring the effectiveness of pressing in soccer. In: Workshop on Machine Learning and Data Mining for Sports Analytics , -.
Journal of Sports Science and Medicine Moura F.A., Martins L.E.B., Anido R., de O., de Barros R.M.L., Cunha S.A. (2012) Quantitative analysis of Brazilian football players’ organisation on the pitch. Sports Biomechanics 11, 85-96.  Crossref
Journal of Sports Science and Medicine Navarro, F.J.F. (2018) Analysis of Styles of Play in Soccer and Their Effectiveness. PhD Thesis, Universidad de Granada.
Journal of Sports Science and Medicine Power, P., Ruiz, H., Wei, X. and Lucey, P. (2017) Not All Passes Are Created Equal: Objectively Measuring the Risk and Reward of Passes in Soccer from Tracking Data. In: Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. KDD '17. New York, NY, USA: Association for Computing Machinery. 1605-1613.  Crossref
Journal of Sports Science and Medicine Rathke A. (2017) An examination of expected goals and shot efficiency in soccer. Journal of Human Sport and Exercise 12.  Crossref
Journal of Sports Science and Medicine Rein R., Raabe D., Memmert D. (2017) “Which pass is better?” Novel approaches to assess passing effectiveness in elite soccer. Human Movement Science 55, 172-181.  Crossref
Journal of Sports Science and Medicine Szczepański Ł., McHale I. (2016) Beyond completion rate: evaluating the passing ability of footballers. Journal of the Royal Statistical Society: Series A (Statistics in Society) 179, 513-533.  Crossref
Journal of Sports Science and Medicine Tenga A., Holme I., Ronglan L.T., Bahr R. (2010) Effect of playing tactics on goal scoring in Norwegian professional soccer. Journal of Sports Sciences 28, 237-244.  Crossref
Journal of Sports Science and Medicine Vilar L., Araujo D., Davids K., Bar-Yam Y. (2013) Science of winning soccer: Emergent pattern-forming dynamics in association football. Journal of Systems Science & Complexity 26, 73-84.  Crossref
Journal of Sports Science and Medicine Vogelbein M., Nopp S., Hökelmann A. (2014) Defensive transition in soccer - are prompt possession regains a measure of success? A quantitative analysis of German Fußball-Bundesliga 2010/2011. Journal of Sports Sciences 32, 1076-1083.  Crossref
Journal of Sports Science and Medicine Weineck J. (2007) Optimales Training : Leistungsphysiologische Trainingslehre Unter Besonderer Berücksichtigung Des Kinder- Und Jugendtrainings. 15. Aufl. Spitta-Verl.
Journal of Sports Science and Medicine Welch M., Schaerf T.M., Murphy A. (2021) Collective states and their transitions in football. Plos One 16, 1-20.  Crossref
 
 
 
Home Issues About Authors
Contact Current Editorial board Authors instructions
Email alerts In Press Mission For Reviewers
Archive Scope
Supplements Statistics
Most Read Articles
  Most Cited Articles
 
  
 
JSSM | Copyright 2001-2024 | All rights reserved. | LEGAL NOTICES | Publisher

It is forbidden the total or partial reproduction of this web site and the published materials, the treatment of its database, any kind of transition and for any means, either electronic, mechanic or other methods, without the previous written permission of the JSSM.

This work is licensed under a Creative Commons License Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.