Sports Medicine, 2012, 42(1), 1-10
The role of ecological dynamics in analysing performance in team sports
Luís Vilar, Duarte Araújo, Keith Davids & Chris Button
Traditional Approaches to Performance Analysis
In team games, notational techniques are used to audit the behaviours of performers during different sub-phases of play. The aim of performance analysis is to provide accurate, augmented information to practitioners to improve future team performance.[1] Practitioners and researchers have revealed a significant interest in the methods of notational analysis because it has successfully described the performance tendencies of players and teams, and strengths and weaknesses in specific performance situations in a range of sports (e.g. playing long or spreading wide during transition phases in association football). This methodology has an important role in creating awareness among players and coaches of how individual players can influence team patterns. Since its initial foundations, a large number of descriptive studies for a variety of sports have been conducted, oftentimes, investigating differences in the action frequencies of players and teams to characterize relationships with successful and unsuccessful performance.[1,2] While this initial approach has provided a valuable foundation to describe performance in team sports, an emerging criticism of performance analysis methods by some sport scientists concerns the need for theoretical principles and empirical data that can help to better explain how successful performance outcomes are achieved.[3]
Sports performance can be measured by scoring indicators (e.g. goals, baskets, winning shots, errors, the ratios of winners to errors and goals to shots) or performance indicators (e.g. turnovers, tackles, passes/ball possession, etc.).[4] These variables are typically recorded in a discrete sequential fashion to operationalize sports behaviours. Each action in a number of games is recorded in a ‘who(did)–what–where–when’ sequence, together with any associated outcomes (i.e. winner, error or neutral outcome).[1] An important criticism of notational analysis research is that it has been somewhat reductionist,[3] typically omitting reference to the ‘why’ and ‘how’ of performance that underlie the structure of recorded behaviours that would define their functional utility.[1] Additionally, the discrete, sequential emphasis on categorizing performance statistics, has led notational analysts to disregard the dependence of a single observation on previous actions (i.e. ‘conditioned coupling’ in performance dynamics).
To maximize the use of notational analysis by coaches, its indicators need to be highly correlated with successful performance[4,5] but, additionally, the relations between a performance behaviour and a specific performance outcome identified by notation analysts needs to be better understood.[1] ,For example, one of the most intriguing indicators of team performance in association football has been ‘style of play’ (e.g. ‘direct play’ or ‘possession play’), measured by the number of passes a team takes to score a goal.[6-14] Although experienced practitioners have emphasized the use of longer passing sequences as a means to score goals, notational analysts have reported that the strike ratio of goals from shots is better for ‘direct play’ than for ‘possession play’.[6] However, the presumed correspondence of this finding to competitive outcomes remains unclear because of a lack of a theoretical understanding of how to interpret the data.
This discrepancy exemplifies how notational analysis methods typically fail to provide a meaningful understanding of successful performance in team games and highlights the need for these methods to be complemented with a sound theoretical rationale of performance behaviours. In this article, we aim to build on Glazier’s[3] recent promotion of dynamical systems theory to explain sport performance by combining these theoretical insights with ideas from ecological psychology, in order to better understand how information from the performance environment constrains the adaptive behaviours of players. Ecological dynamics is a viable framework for studying behaviours in team games because it recognizes the ‘degeneracy’ (inherent adaptive flexibility in achieving successful performance outcomes) of neurobiological systems (i.e. athletes), including social neurobiological systems (sports teams). Its principles can clearly explain how the same successful performance outcomes can emerge from different movement or tactical patterns (a phenomenon also known as motor equivalence).[15]
An Ecological Dynamics Account of Performance
In studies of complex neurobiological systems, ideas from ecological dynamics have provided a powerful theoretical explanation of behaviours. Ecological dynamics proposes performer-environment relations as the relevant scale of analysis for understanding sport performance.[16] Functional patterns of coordinated behaviour emerge through a process of self-organization from performers’ interactions with each other under specific task and environmental constraints.[17,18] In complex social neurobiological systems, self-organization is the fundamental principle for explaining how order emerges amongst different components.[19] Ecological dynamics analyses of team sports have attempted to explain how the interactions between players and information from the performance environment constrains the emergence of patterns of stability (i.e. coordination between performers), variability (loss of coordination between performers) and symmetry-breaking in organizational states (i.e. how new patterns of coordination emerge in performance) in such systems. This is precisely what sport scientists and coaches need to understand in analyses of team game performance.[17,20,21]
The emergent coordination patterns in team sports are channelled by surrounding constraints (e.g. location of opponents, team-mates, the ball and goals) that shape the emergence of playing configurations.[22] Constraints provide information that affords opportunities for players to act.[23] Such opportunities for action are shaped by both the performance environment and the characteristics of each performer (e.g. prospective and retrospective memories, such as remembering what to do when, or remembering what may happen when).[17] For example, an opportunity to score a goal in football may emerge between the ability of the performer to shoot the ball (individual constraints) and the distance to the goal (task constraints). In this way, successful performance in sport is grounded on the performer’s ability to attend to the relevant informational variables that are needed to regulate their decisions and actions.[24] Moreover, performers are also able to identify relations between other performers (e.g. team-mates and opponents) and key environmental objects (e.g. the ball and goal in team games) that can constrain their own behaviours.[25] By perceiving opportunities for others to act, performers make use of environmental information to coordinate their actions with others.
Because of the complex spatiotemporal relations among performers that characterize team sports, performance constraints can change instantaneously. Opportunities to act may appear and disappear quickly, leading to fluctuations in the organizational states of games (e.g. increased variability in the way that attackers and defenders coordinate their actions).[26] When these fluctuations are powerful enough to break the existing balance between performers (i.e. if the equilibrium between attacking and defending players is successfully destabilized), a symmetry-breaking process occurs. That is, a previously stable state of the game transits to a new dynamic state of organization (e.g. an attacker dribbles past a first defender, inducing a second defender to cover, leading to a change in the structure of the defending team).[23]
Interpersonal Coordination in Team Sports
To investigate interpersonal coordination in team games, the ecological dynamic studies of performance have undertaken analyses of the emergent patterns of coordination in attacker-defender sub-systems (e.g. one vs one or two vs one) as a basic unit of investigation.[18,19] In football, for example, attacker-defender dyadic sub-systems are continuously forged and broken as constraints change during the players’ efforts to score goals or regain ball possession.[27] The aim of each attacker is to use his/her coupling with an immediate marking defender in the most beneficial way for the team. For example, by (i) avoiding attentions of the nearest defender to promote an instability in the dyadic system and, therefore, in the defensive organization; or (ii) ‘fixing’ an immediate opponent in a stable dyad so that other dyads may attain criticality (allowing other attackers to exploit the attacking potential in other one versus one sub-systems, without a covering defender). In dyadic sub-systems, each defender seeks to counteract the movements of immediate attackers, in order to maintain system symmetry (to prevent the immediate attacker from gaining a positional advantage that allows him/her to threaten the goal).[27,28] However, with decreasing interpersonal distances between opposing players, a state of ‘criticality’ in one or more dyadic systems may be attained.[29] At this point, a sudden change in the structural organization of the system(s) towards one of two possible states may be about to occur, displaying (i) an advantage for an attacker (e.g. the defenders may not be able to balance the attackers’ actions and the attackers may move past their immediate opponents); or (ii) an advantage for a defender (e.g. the attacker is not able to break system symmetry and a defender may intercept the ball, reversing the attacker-defender roles). Either way, system stability during team games can be destroyed and a new pattern of coordination in the team game can suddenly emerge through a process known as a symmetry breaking.[18,28-31] This emergent process may lead to changes in the structural organization of team games, captured by the appearance and disappearance of offensive and defensive patterns of play, which in turn can transform the whole game context.[22]
A key task in this type of theoretical performance analysis is to identify collective variables that capture system organization and its changes over time. Some previous research in basketball has considered the spatial trajectories of performers as nonlinear coupled oscillators, suggesting ‘relative phase’ as a candidate for a collective variable. This work has proposed that the dynamics of relative phase might permit the quantitative expression of coordination processes between dyadic system performers in team games.[32] When both performers in a dyad move forwards and backwards (or from side to side) at the same time, an ‘in-phase’ (0º) coupling between players might be identified. On the other hand, an ‘anti-phase’ (180º) mode of coordination emerges when one performer is moving in an opposite direction to another player at the same time.[33] Since relative phase may be expressed in terms of circular statistics, −360º and −720º denote the same phase relation (0º). Analyses of longitudinal (i.e. forward and backward) movement displacement trajectories of attackers and defenders in team games have revealed strong attractions to in-phase patterns, and a meta-stable distribution (with no preferential mode of coordination) in lateral movement displacement.[32]
Another important task is to gain understanding of how coordination in dyadic systems might be constrained by the location of key objects in the field of play, such as the goal area and the ball that might shape the actions of team sports performers.[18] The next section illustrates how research on one versus one sub-phases of team sports has conceptualized how performers might use information from the environment to score/prevent goals in competitive performance. As a task vehicle, we discuss extensive work on performance in futsal (a type of five-a-side Association Football played on an indoor court) using relative phase analysis of movement displacements between immediate opponents.
Experimental Observations of Emergent Patterns of Coordination in Futsal
Previous research on one versus one sub-phases in basketball performance has suggested that a symmetry-breaking in the distances of immediate attackers and defenders to the basket can precipitate a scoring opportunity.[18,19] In contrast, research on futsal has revealed how attackers often create shooting opportunities, not by transiting beyond a defender, but by simply promoting a misalignment in a defender’s positioning between the attacker and the goal. Figure 1 depicts data from the ten futsal games in the 2009 Lusophony Games held in Lisbon, Portugal, in which 13 sequences of play ending in a goal scored were randomly selected. In each sequence, the interactions between the four outfield attackers and the nearest defender (n = 52 trials) with the goal were analysed. We considered the positioning of the players relative to the centre of the goal because it represents the mean vector of all possible ball displacement trajectories for a goal to be scored (i.e. between one corner of the goal and the other). Results showed how interpersonal coordination processes in each attacker-defender dyad emerged and were constrained by distances (figure 1a, c and e) and angles (figure 1b, d and f) of the performers to the goal. While the upper and middle panels represent one trial of the attacker who scored the goal (in one exemplar situation), the bottom panels display the frequencies of modes of coordination of all trials in each frame of each play.
Results demonstrated how in-phase patterns of coordination emerged from changes in the values of attackers and defenders distances to goal (figure 1e, 46% of total time in 0º bin, i.e. within 0º and 29º phase limits) ((Author: please define ‘bin’))) and angles to goal (figure 1f, 49% of total time in 0º bin). In futsal games attackers try to break symmetry not only by displacing laterally to increase the angle to the goal relative to a marking defender’s position, but also by decreasing their distance to the goal relative to the defender. Alternatively, defenders tried to maintain system symmetry by placing themselves closer to the goal, between it and the immediate attacker. Data suggested that goals were precipitated by more than one symmetry-breaking process involving the relative distances of attackers and defenders to the goal (figures 1c: transitions between 0º and −360º at 23 sec, and −360º to −720º at 33 sec) and their angles to goal (figure 1d, transitions between 0º and 360º at 20 sec, and 360º to 0º at 38 sec). This important observation may imply that the ability to destabilize or (re)stabilize these sub-systems can be considered a hallmark characteristic of skilled performance in competitive team games.[27,28] As sequences of play evolved towards the goal, the durations of stable system states decreased, providing attackers with more frequent opportunities to score goals. Importantly, it was only when symmetry-breaking processes emerged near to the goal, and when the defenders did not have the collective ability to re-establish dyadic system stability, that a significant goal opportunity presented itself.
In order to understand successful and unsuccessful performance outcomes, we analysed the emergence of this system’s dynamics in plays ending in a goal (n = 30), a goalkeeper’s save (n = 30) and a defender’s interception (n = 30), at four discrete moments, sequentially ordered, that led to a shot: (i) the assistant attacker’s ball reception and (ii) pass; and (iii) the shooter’s ball reception and (iv) shot on goal. Figure 2 shows how the pattern forming dynamics of an attacker with the ball and a marking defender were constrained by the information from several key sources including: (i) the defender’s angle to the goal and the attacker (inner product of the defender’s vector to the centre of the goal, and the defender’s vector to the attacker) [figure 2a]; (ii) the relative distance to the goal (difference between the attacker’s and the defender’s distances to the centre of the goal) [figure 2b]; (iii) the interpersonal distance between the attacker and defender (figure 2c); and (iv) the relative velocity between the attacker and defender (figure 2d).
Data revealed that when a goal was scored, compared to when the ball was intercepted, a number of features were observed in the interpersonal coordination between defenders and attackers. For example, the defender’s angle to the goal and to the attacker tended to decrease, suggesting that the attacker promoted a misalignment in the defender’s position relative to the goal and the immediate attacker. Although a slight misalignment of the defender (150º) was also observed in plays ending in a defender’s interception, there was enough time for the defender to move to intercept the trajectory of the ball (figure 3a). Results also suggested that the attacker was able to move to the same distance to the goal as the defender, making it difficult for the defender to intercept the trajectory of the ball to the goal. Conversely, the closer the defender was to the goal and the further attacker was to the goal, the more frequent were the opportunities to prevent a goal from being scored (figure 3b). When an attacker scored a goal, he/she((Author: change to he/she and him or his/her elsewhere for consistency?)) was able to maintain a significantly larger distance value between him/her and a marking defender. This distance value was not only significantly larger when the attacker had the ball in his possession, but also while he moved to provide a passing opportunity for his/her attacking team-mate (figure 3c). Finally, to score a goal, the attacker needed to maintain a high velocity both before and after receiving the ball. This strategy prevented the defender from getting close enough to the attacker to intercept the ball’s trajectory. Conversely, the defender needed to increase his velocity (than that of the attackers) as soon as possible to meet the required velocity necessary to intercept the ball’s trajectory (figure 3d).[1] Additionally, even when the defender was not able to intercept the ball’s trajectory, our findings suggested that he/she should continue to pressurize the attacker, since these actions constrained the attacker to shoot sooner than needed, affording the goalkeeper enough time to intercept the ball (see lines with squares in figure 3d. This observation is also illustrated in the remaining panels of Figure 3, since the mean values of all variables in plays ending in a goalkeeper’s save are in between the mean value of plays ending in a goal and those ending in a defender’s interception.
Figure 1. The constraint of goal location on coordination processes in dyadic systems presented in decomposed format:left column–distances of each player to the centre of the goal; right column – angles of each player to the centre of the goal. (a) and (b) Exemplar data from attacker five [A5] and nearest defender [D]. (c) and (d) Dynamics of the relative phase of the exemplar data from A5 and nearest D. (e) and (f) Frequency histograms of the relative phases of all A–D dyadic systems (n = 52).[34]
Figure 2. Illustration of the different variables analysed in this investigation. (a) [1] Nearest defender’s (Ds) angle to the attacker (A) and the centre of the goal. (b) Relative distance to the goal computed by the difference between the A distance to the centre of the goal [2] and the nearest Ds distance to the centre of the goal [3]. (c) Interpersonal distance [4] between the A and the nearest D. (d) Relative velocity computed by the difference between the A velocity [5] and the nearest Ds velocity [6].
Figure 3. Mean values and SDs of the defender’s (Ds) angles to the goal and the attacker [A]. (a) The relative distance to the goal. (b) The interpersonal distance. (c) The relative velocity. (d) The three types of performance outcomes at the four instances prior to the shot at goal (SG): moment of ball reception by the assisting receiving (AR) player, moment of pass initiation by the assisting player (AP), moment of ball reception by the shooting receiving (SR) player and moment of SG.[35] GKs = goal keepers.
Conclusions and Implications
In this article we have argued why ecological dynamics is a theoretical explanation of performance analysis that develops understanding of the emergence of successful and unsuccessful patterns of play in team sports. Data from analyses of interpersonal spatial and temporal interactions have revealed how performance of individuals emerged from self-organized processes under the constraining influence of the locations of their opponents, the ball and the goal. Ecological dynamics research has shown how attackers try to break symmetry with their nearest opponents, as defenders seek to maintain system symmetry by remaining between their own goal and the immediate attacker. To score goals, attackers should create instabilities in a dyadic system near the goal by creating space to shoot the ball before the defender gets close enough to intercept the ball’s trajectory. These theoretical ideas from ecological dynamics suggest how ubiquitous ball possession data in football (usually expressed vaguely as a percentage of total time in possession of the ball by a team) can be interpreted relative to successful performance outcomes. In this article we suggest that ball possession per se is not necessarily a condition of successful performance in team games. Although still needing further empirical demonstration, ideas from ecological dynamics clearly reveal that the aim of ball possession and associated styles of play in team sports is to seek to create system transitions. These abrupt transitions can emerge from opportunities to break dyadic sub-system stability locally with immediate opponents, leading to changes in the large scale defensive organization and allowing goal opportunities to emerge. These symmetry-breaking processes can emerge from 2 or 20 seconds of ball possession, or with a style of play involving 2 or 20 passes.
We have proposed how ecological dynamics might provide traditional approaches to performance analysis with a sound theoretical rationale in order to substantiate insights into successful and unsuccessful performance outcomes in team sports. By considering the functionality of individual actions, this theoretical rationale might enhance the current operational nature of performance analysis that typically correlates performance behaviours to performance outcome through use of discrete, descriptive statistics. To enhance the validity of theoretical interpretations, analysts need to move beyond merely documenting ‘performance statistics’ in order to study the emergent interactions between players, in key areas of the field, which underpin success in team sports.
Ecological dynamics also has major implications for the design of representative training tasks in team sports. The key informational constraints that afford opportunities for performers to stabilize or destabilize sub-systems of play can be simulated (i.e. represented) in training tasks, allowing them to become better attuned to the information variables that constrain performance, and to functionally couple information and movement during practice.[36] Major implications of these ideas also exist for development programmes in team sports. Instead of decomposing tasks as repetition drills for learners, coaches should simplify tasks (e.g. by reducing the numbers in teams during small-sided practice games (e.g. four vs four, seven vs seven or six vs four) to encourage adaptive movement behaviours and to simulate the essential information available to performers). By unveiling the influence of interacting task constraints on the emergent self-organized behaviours of players during performance, ecological dynamics reveals itself as a powerful tool for both researchers and practitioners in sport performance analysis.
Acknowledgements
Luís Vilar was supported by a financial grant from the Portuguese Foundation for Science and Technology (SFRH/BD/43251/2008). The authors would like to acknowledge the three anonymous reviewers for their insightful and valuable comments on earlier versions of the manuscript. The authors have no conflicts of interest to declare that are directly relevant to the content of this article. No funding was received by Duarte Araújo, Keith Davids and Chris Button to assist them in the preparation of this article.
References
[1] McGarry T. Applied and theoretical perspectives of performance analysis in sport: scientific issues and challenges. Int J Perform Anal Sport 2009; 9: 128-40
[2] Hughes M, Franks I. Notational analysis: a review of the literature. In: Hughes M, Franks I, editors. Notational analysis of sport. 2nd ed. London: Routledge, 2004; 59-106
[3] Glazier P. Game, set and match? Substantive issues and future direction in performance analysis. Sports Med 2010; 40 (8): 625-34
[4] Hughes M, Bartlett R. The use of performance indicators in performance analysis. J Sports Sci 2002; 20: 739-54
[5] Lago C, Martin R. Determinants of possession of the ball in soccer. J Sports Sci 2007; 25 (9): 969-74
[6] Hughes M, Franks I. Analysis of passing sequences, shots and goals in soccer. J Sports Sci 2005; 23 (5): 509-14
[7] Reep C, Benjamin B. Skill and chance in association football. J R Stat Soc 1968; 131 (4): 581-5
[8] Reep C, Pollard R, Benjamin B. Skill and chance in ball games. J R Stat Soc 1971; A (134): 623-9
[9] Franks I, Goodman D, Miller G. Human factors in sport systems: an empirical investigation of events in team games. Proceedings of the Human Factors and Ergonomics Society 27th Annual Meeting; 1983 Oct 10-14. Hum Fac Soc 1983; 27 (5) 383-6.
[10] Franks I, Partridge D, Nagelkerke P. World Cup 90: a computer assisted technical analysis of team performance; 1990.
[11] Hughes M, Robertson K, Nicholson A. An analysis of the 1984 World Cup of Association Football. In: Reilly T, Lees A, Davids K, Murphy W, editors. Science and football. London: E & FN Spon, 1988: 363-7
[12] Partridge D, Franks I. A detailed analysis of crossing opportunities from the 1986 World Cup (part I). Soccer J 1989 May-Jun: 47-50
[13] Partridge D, Franks I. A detailed analysis of crossing opportunities from the 1986 World Cup (part II). Soccer J 1989 Jun-Jul: 45-8
[14] Grehaigne J. Systemic approach and soccer. In: Hughes M, editor. Notation of sport III. Cardiff: Centre for Performance Analysis, UWIC, 1999; 1-8
[15] Scholz JP, Schoner G, Latash ML. Identifying the control structure of multijoint coordination during pistol shooting. Exp Brain Res 2000 Dec; 135 (3): 382-404
[16] Davids K, Araújo D. The concept of ‘Organismic Asymmetry’ in sport science. J Sci Med Sport 2010; 13 (6): 663-40
[17] Araújo D, Davids K, Hristovski R. The ecological dynamics of decision making in sport. Psychol Sport Exerc 2006; 7: 653-76
[18] Araújo D, Davids K, Bennett S, et al. Emergence of sport skills under constraints. In: Williams AM, Hodges NJ, editors. Skill acquisition in sport: research, theory and practice. London: Routledge, Taylor & Francis, 2004; 409-33
[19] Davids K, Button C, Araújo D, et al. Movement models from sports provide representative task constraints for studying adaptive behavior in human movement systems. Adapt Behav 2006; 14 (1): 73-95
[20] Davids K, Handford C, Williams M. The natural physical alternative to cognitive theories of motor behaviour: an invitation for interdisciplinary research in sports science? J Sports Sci 1994; 12 (6): 495-528
[21] Handford C, Davids K, Bennett S, et al. Skill acquisition in sport: some applications of an evolving practice ecology. J Sports Sci 1997; 15: 621-40
[22] Gréhaigne J, Bouthier D, David B. Dynamic-system analysis of opponent relationships in collective actions in soccer. J Sports Sci 1997; 15 (2): 137-49
[23] Davids K, Glazier P, Araujo D, et al. Movement systems as dynamical systems: the functional role of variability and its implications for sports medicine. Sports Med 2003; 33 (4): 245-60
[24] Hristovski R, Davids K, Araújo D, et al. How boxers decide to punch a target: emergent behavior in nonlinear dynamical movement systems [combat sports special issue]. J Sports Sci Med 2006; 60-73
[25] Richardson M, Marsh K, Baron R. Judging and actualizing intrapersonal and interpersonal affordances. J Exp Psychol Human 2007; 33 (4): 845-59
[26] Araújo D, Davids K. Ecological approaches to cognition and action in sport and exercise: ask not only what you do, but where you do it. Int J Sport Psychol 2009; 40 (1): 5-37
[27] McGarry T, Anderson D, Wallace S, et al. Sport competition as a dynamical self-organizing system. J Sports Sci 2002; 20: 771-81
[28] Davids K, Araújo D, Shuttleworth R. Applications of dynamical systems theory to football. In: Reilly T, Cabri J, Araújo D, editors. Science and football V. London: Routledge, 2005; 547-60
[29] Passos P, Araujo D, Davids K, et al. Information-governing dynamics of attacker-defender interactions in youth rugby union. J Sports Sci 2008; 26 (13): 1421-9
[30] Passos P, Araújo D, Davids K, et al. Interpersonal pattern dynamics and adaptive behavior in multiagent neurobiological systems: conceptual model and data. J Motor Behav 2009; 41 (5): 445-59
[31] Duarte R, Araújo D, Gazimba V, et al. The ecological dynamics of 1v1 sub-phases in Association Football. Open Sports Sci J 2010; 3: 16-8
[32] Bourbousson J, Seve C, McGarry T. Space-time coordination dynamics in basketball: part 1. Intra- and inter-couplings among player dyads. J Sports Sci 2010; 28 (3): 339-47
[33] Palut Y, Zanone P. A dynamical analysis of tennis: concepts and data. J Sports Sci 2005; 23 (10): 1021-32
[34] 34. Vilar, L., Araújo, D., Davids, K., & Travassos, B. (in press). Constraints on competitive performance of attacker-defender dyads in team sports. Journal of Sports Sciences. doi: 10.1080/02640414.2011.627942
[35] Vilar L, Araújo D, Davids K, et al. Decision-making in attacker-defender dyadic systems in Futsal. In: Serpa S, editor. 13th European Congress of Sport Psychology; 2011; Madeira; 2011.
[36] Davids K, Renshaw I, Glazier P. Movement models from sports reveal fundamental insights into coordination processes. Exerc Sport Sci Rev 2005; 33 (1): 36-42