ISBN:

Transcription

1 Chapter 12 K.A.P.M. Lemmink & W.G.P. Frencken (2009) Physiological and tactical match analyses in ball team sports; New perspectives. In: Aportaciones teóricas y practices para el baloncesto del futuro (Eds. A. Lorenzo, S.J. Ibáñez & E. Ortega). Wanceulen Editorial Deportiva, SL, Sevilla, Spain. ISBN:

2 Physiological and tactical match analyses in ball team sports; New perspectives Koen A.P.M. Lemmink 1,2 Wouter G.P. Frencken 1 1 Center for Human Movement Sciences, University Medical Center Groningen, University of Groningen, the Netherlands 2 School of Sports Studies, Hanze University of Applied Sciences, Groningen, the Netherlands Match analysis is the objective recording and examination of behavioural events of one or more players during competition or training. The primary goal of match analysis is to provide information to coaches and players about team and/or player performance in order to plan subsequent practices to improve performance or to support preparation for the next match (Hughes & Franks, 2008; Carling et al., 2009). Depending on the goal of match analysis, information on performance can be obtained from a biomechanical, technical, physiological or tactical perspective. This chapter will focus on new perspectives in physiological and tactical performance analysis. Physiological performance Motion analysis focuses on the frequency, duration and exercise intensities of the different activities during competition and training to quantify the specific physiological requirements of the sport. Match activities are coded according to the intensity of the movement activities, such as standing, walking, jogging, running, and sprinting. Most often, video-based notation techniques are used to gather information on distances en exercise intensities. However, these techniques are time-consuming, not always accurate and limited to the analysis of one single player at a time. In recent years, technological innovations, such as (automatic) video-based tracking and GPS like technology, have led to new possibilities for motion analysis in ball team sports. Positional data of all players with high frequency is becoming available

3 in the context of different ball team sports, such as soccer, field hockey, basketball, rugby, and handball. Video-based systems for tracking players, such as PROZONE, AMISCO Pro, or SportVU, generally require the installation of multiple cameras to cover the whole field and a calibration procedure allowing calculation of player positions from the camera viewpoints. Mostly, an operator is needed to improve the accuracy of tracking and data are not real-time available. Electronic tracking systems, such as Local Position Measurement (LPM) or GPSports, require tagging players electronically by way of antennas and transponders to track their movements by means of radio frequent signals picked up by receivers around the field or satellites. The LPM Inmotio system can be used real-time and is highly accurate. EXAMPLE 1 Motion analysis Futsal Futsal, i.e. five-a-side indoor soccer, is an intermittent sport with increasing popularity all over the world. It is played on a 40 x 20 m court with 3 x 2 m goals for two 20-min periods. A team compromises five players, four field players and a goalkeeper. To analyse the activity profile and physiological demands of futsal two training matches of the Dutch national team (n=14; age 26.3±3.2 years, length 180±6 cm, weight 76.3±6.3 kg) were monitored using Local Position Measurements (LPM, Inmotio Object Tracking BV, Amsterdam, the Netherlands). Positional data was synchronized with heart rate data (Polar, Kempele, Finland). The players were equipped with a vest containing a transponder. Ten beacons surrounded an indoor field recording radio frequent signals from the players at 62Hz. Furthermore, one dome camera above each playing half was used as to follow the progress of play and to interpret the LPM data afterwards. Following Barbero- Alvarez et al. (2008), six categories of match activities were used: standing ( km h -1 ), walking ( km h -1 ), jogging ( km h -1 ), medium-intensity running ( km h -1 ), high-intensity running ( km h -1 ) and sprinting ( 25 km h -1 ). Mean distance covered during a match was 4868 m±741 m, mean playing time 39.17±6.08 min, and mean relative distance, i.e. distance covered per playing minute, m min -1 ±9.1. Mean distances covered per match activity (m) are presented in Figure 1.

4 ,6 % Distance (m) ,5 % ,1% 9,9 % 7,0 % Standing Walking Jogging MI Running HI Running Sprinting Distance (m) SD Minimum Maximum ,8 % Figure 1. Mean distances and percentages covered in the different match activity categories: MI=Medium Intensity, HI=High Intensity. Distance (m min -1 ) * Distance (m) First half Second half Distance (m min -1 ) Distance (m) * 1500 Figure 2. Mean relative distance (m min -1 ) and total distance (m) covered during the first and second halves: * p<.05.

5 There were no differences between mean total distances covered during the first and second halves; 2481±398 m and 2387±478 m, respectively. However, relative distance decreased significantly from the first to the second halves; from 127.2±7.8 m min -1 to 121.2±11.5 m min -1 (p<.05). This is illustrated in Figure 2. Per match activity category there were no differences in mean distance covered and percentage of total distance covered between the first and second halves (Figure 3). Mean heart rate in two training matches was 170±9 beats min -1 with a range from 154 to 185. There were no differences in mean heart rate between the first and second halves. 120 Distance (%) Standing Walking Jogging MI Running HI Running Sprinting First half Second half Figure 3. Mean percentages of distance covered in the different match activity categories during the first and second halves: MI=Moderate Intensity, HI=High Intensity. Data with high frequency and accuracy, such as collected with LPM in the futsal example, open up possibilities to develop new methods to calculate physiological load based on players speed, accelerations, decelerations and directional changes. Momentarily, a study is carried out to calculate training load based on speed and acceleration/deceleration profiles of professional soccer players.

6 Tactical performance of on-the-ball players Notational analysis is a method to create a permanent record of the on-the-ball actions of players within a game through hand-based or computerised systems often using video technology. Basic systems simply classify the actions of the players (Figure 4). More sophisticated systems measure sequence of actions, time of actions and the position of the actions on the field which allows temporal and spatial analysis of the data. Other systems have tried to define actions, e.g., passes, in terms of successful or unsuccessful or to rate the quality of actions on a 3 or 5-point scale. Combined with video clips, these analyses are used by coaches to provide the team or individual players with feedback on their tactical behaviour during match-play and to prepare for future opponents. Nowadays, several computerised notational systems can be used real-time and have features, i.e., video clips or statistics, that allow coaches to use this information during half-time or to instruct individual players before entering the game. Figure 4 Print screen of a notational system for basketball (Track 1.1., University of Groningen, the Netherlands) Next to the real-time or day-after use of a single match, notational systems can be used to analyse a series of matches during tournaments, such as European or World Championships, and competitions. Differences between successful and unsuccessful

7 teams in action profiles can shed light on key performance indicators from a tactical perspective. EXAMPLE 2 Notational analysis Dutch professional soccer league All 308 matches of the Dutch professional soccer league (highest level) were analyzed afterwards using the computerized video notation system Effectivity in Action (ORTEC, Gouda, the Netherlands). Trained analysts notated the matches by classifying the actions of the player holding the ball. The analysts were trained and had to perform an exam by analyzing a standardized match. The results of the analysis were only allowed to deviate to a small extent when compared to the analyses of a group of experts, providing the notation system with face validity. The notation system required two analysts per team, for data classification and data entry respectively. The analysts were positioned in front of a big TV screen and a touch screen PC to enter the data. A senior analyst supervised the incoming data from both teams. After supervision, the data were stored in a computer database for (real-time) data processing. Every data entry was provided with a time tag and stored chronologically. The database consisted of match information, general statistics and all actions of the players holding the ball. Every single action was categorized and subsequently judged qualitatively. Six key performance indicators were taken into account; 1) goal attempts, 2) passes, 3) interceptions/duels, 4) set plays 5) fouls/off sides and 6) goal keepers actions. The quality of every single action was rated on a three point scale, i.e. low quality, moderate quality or high quality. The rating scheme was developed together with a team of expert soccer coaches. In total, player actions were notated with an average of 674 (range ) player actions per team per match. In total 870 goals were scores out of 8464 goal attempts (10.3%). Without penalty kicks, shooting within the penalty area was the most effective way of scoring goals (15.5%) followed by headers (12.7%), free kicks (5.1%) and shots outside the penalty area (3.4%). Most goals were scored in the second half and more specific in the last 15 minutes of the matches (25.4%). In contrast, only 10.5% of the goals were scores in the first 15 minutes (Figure 5).

8 Number of goals Figure 5 Number of goals per 15-min period during the matches of the Dutch professional soccer league Periods Teams were classified as either successful (place 1-9 on the final ranking list) or unsuccessful (place on the final ranking list). Successful teams had more goal attempts and more successful goal attempts within the penalty area and with the head. They were also more effective in scoring goals out of these goal attempts. The superiority of the higher ranked teams in goals and goal attempts is illustrated in figure (N Goals/attempts)x100 N Attempts/match N Final ranking Figure 6 Number of goal attempts/match and ((number of goals/attempts) x100) in relation to the final ranking of the teams in the Dutch professional soccer league

9 Successful teams also outscored the unsuccessful teams in passing behaviour, i.e. more passes backwards/ sideways, forward passes and crosses but also a higher quality of passes. The latter is illustrated in figure 7 and 8. The higher the final ranking of a team, the more (moderate) sideways/backwards and forward passes they produced (Figure 7). There was also a trend that better ranked teams produced more high quality passes and less low quality passes (Figure 8). Passing backw/sideways Moderate passing backw/sideways Passing forward Moderate passing forward Frequency Final ranking Figure 7 Number of (moderate) passes backward/sideways and forward per match in relation to the final ranking of the teams in the Dutch professional soccer league Large amounts of sophisticated notational data, as in the Dutch league example, open up to new approaches. For example, temporal (T-pattern) analysis and network approaches seem promising to expand our knowledge on tactical performance. Although these notational systems have improved over time, they still have certain limitations, especially from a tactical point of view. For example, information of position of the actions lack accuracy and, due to a single camera viewpoint, only onthe-ball actions of individual players are monitored properly.

10 HQ Passing backw/sideways LQ Passing backw/sideways HQ Passing forward LQ Passing forward Frequency Final ranking Figure 8 Number of high quality (HQ) and low quality (LQ) passes backward/sideways and forward per match in relation to the final ranking of the teams in the Dutch professional soccer league Tactical performance of interacting players As shown previously, notational analysis in ball team sports primarily focuses on performance of individual players holding the ball. However, it can be argued that the behaviour of an individual player in a match is brought about by interactions with his or her environment. In other words, teammates, opponents, referees and others influence all individual players in one way or another. So, the fact that sports like futsal, basketball and soccer are invasion games implies that they are of complex nature in which technical, physical, mental and tactical components are interrelated. This aspect is often ignored in current scientific literature, because in most studies the dynamics of a match are not taken into account. Instead, series of discrete events of individual players are analyzed. One way to deal with these aforementioned issues is to analyze positional data of the players. This is rapidly becoming common practice, as more and more quick and accurate methods are available to monitor player positions throughout a match or training. For an overview of these methods, see Carling et al. (2005, 2009). Subsequent positions of individual players, or a changes in position, reflects the interaction of a player with its environment. In these positional changes, all

11 components are incorporated. The direction is which a player moves reflects his decision to move in that particular direction. Secondly, the speed or acceleration by which a player changes position may represent his physical status. Finally, the interaction between the players is reflected by the configuration of the players on the field. As a match progresses, the player positions change continuously. This means that spatio-temporal patterns emerge. In order to study these spatio-temporal patterns, a different theoretical framework must be adopted, since for example the aforementioned T-patterns mainly focus on temporal patterns. One such framework is dynamical systems theory (DST). This theory was first introduced for research in human behaviour by Kugler (1980). The core of this theory is that the behaviour of the system is brought about by interactions of many subsystems. For example, in humans the muscular-skeletal, nervous and other systems interact to form the human body. The human body is much more complex than the individual subsystems. In sports, this means that the team is more than the sum of all individual players. Several important experiments in mainly bimanual coordination have been performed since then that provide support for this theory (e.g. Kelso, 1985; Schmidt et al., 1990). The first experiments concerned coordination patterns within a single individual. This was extended to coordination patterns between individuals. These studies share two important characteristics, namely that coordination emerged spontaneously as a result of interacting subsystems and that coordination patterns preferably settled into two modes, either inphase or antiphase. The dynamics of these patterns can be modelled and described by a coupled oscillator model, referred to as the HKB-model (Haken et al., 1985). The dependent variable, or collective variable, in this case describes the state of the system. Theoretically, the number of independent variables is unlimited. More independent variables increase the degrees of freedom of a dynamical system. In other words, with increased degrees of freedom, the complexity increases. Finding the collective variable that captures the dynamics of a system is an important scientific challenge. In the last decade, this framework has sporadically found its way into sports performance literature. McGarry et al. (2002) have been instrumental in this. They proposed that interactions between people give rise to team behaviour and may be described as a dynamical system. We proposed earlier that the interactions between players are reflected by positional data of the players. Therefore, positional data of the

12 players can be used to describe the dynamics of a sports contest. Some authors have focused on 1 vs. 1 attacker-defender dyads, in which complexity is reduced, compared to 11 vs. 11 games in soccer. Araújo et al. (2004) analyzed 1 vs. 1 situations in basketball by means of positional data. They calculated the median point of the distance between the athletes to the goal area and the interpersonal distance between attacker and defender, with the former being the collective variable and the latter being a control parameter. Results showed that the attacker fluctuates the direction of attack in front of the defender and the defender countermoves in order to maintain stability. Superiority of the attacker results in dribbling past the defender, whereas superiority of the defender results in maintenance of initial stability. From these data, Araújo et al. (2004) concluded that features of dynamical systems were established in a 1 vs. 1 attacker-defender dyad in basketball. Passos et al. (2006, 2008) performed similar analyses in attacker-defender dyads in rugby. However, they showed that next to interpersonal distance, relative velocity between the attacker and defender was an important control parameter. They also demonstrated that they are intertwined, because at a given interpersonal distance, a high relative speed means superiority for the attacker whereas low relative velocity means superiority for the defender. The implications for sports practice are to encourage players to explore the relative speed that is required to pass a defender. EXAMPLE Araújo et al. (2004) and Passos et al. (2006, 2008) have analyzed discrete 1 vs. 1 attacker defender dyads to identify parameters that determine if the attacker passes the defender. The next step in this process is to assess if the same principles hold for real matches. Our analyses aimed at indentifying similar patterns during small-sided soccer games. We recorded player positions throughout two consecutive 4-minute small-sided games by means of the local position measurement (LPM) system. We subsequently calculated the interpersonal distance between attackers and defenders during the game. We used video analysis to determine the 1 vs. 1 situations. The 1 vs. 1 situations were defined as all moments an attacker tried to dribble past his direct opponent. Results of our study are comparable with results found in basketball (Araújo et al., 2004). If the attacker breaks the symmetry, and the defender is unable to restore symmetry, a goal-scoring opportunity arises. This is demonstrated in figure

13 9. Similar patterns can be seen in 1 vs. 1 situations that are not directly related to goal-scoring opportunities, but that occur elsewhere on the pitch (figure 10) Distance (m) Attacker Defender Time (s) Figure 9. Distance of the attacker and defender to the goal. A goal is scored on t= 9.3 s (black vertical line) Distance (m) Attacker Defender Time (s) Figure 10. Distance of the attacker and defender to the goal during the game. The attacker goes past the defender at t= 10 s

14 A basketball or soccer match includes more than discrete, short term 1 vs. 1 situations. So, the overall team behaviour is more complex. Frencken and Lemmink (2008) have studied small-sided soccer games (4 vs. 4). Two variables that may potentially capture the dynamics were introduced: the centroid position of the teams and the surface area of the teams. The centroid position is the center of a team, that is the ( x, y) of the outfield players. The surface area is the space one team covers. This can be visualized by putting an elastic band around the players of that team. These variables are strongly related to concepts of the Dutch total football philosophy. The centroid position reflects pressing or fore checking. The surface area represents freeing up space when in possession, and closing down space when the ball is lost. Based on this tactical knowledge, an in phase relation was hypothesized for the centroid position and an antiphase relation was hypothesized for surface area. Visual inspection of the data showed that the centroid positions of both teams move in phase as the game evolves. No clear antiphase pattern was seen for the surface area in the corresponding period. Closer examination of the data revealed that when most goals were scored, the centroids exchanged position. So, the attacking team overpowered the defending team. Based on video analysis, it was determined that centroids crossed, when players were quickly moving. Therefore, we argued that the rate of change of the centroid positions is very important in scoring goals. This is similar to findings of Passos et al. (2008), who has shown that relative velocity is a key control parameter in an attacker-defender dyad. This has up to date not been empirically shown in team sports and needs to be established. At the moment, several research groups work on the ideas presented in this chapter in numerous ball team sports, including handball, rugby, soccer and basketball. Future studies should aim at identifying specific properties of a dynamical system. In addition, these analyses must lead to clear guidelines for coaches. Concluding remarks Throughout this chapter, a variety of approaches in analyzing match performance in ball team sports have been discusses, especially with regard to physiological and tactical performance in soccer. New technologies allow for positional data collection with high frequency and accuracy on a regular base during training and matches. For physiological analysis, these data sets open up to defining physiological load of

15 individual players in terms of distance, speed, acceleration, deceleration and directional changes, not only in relation to performance but also in relation to injury risk and reconditioning programs. For the analysis of tactical behaviour, large data sets based on notational analysis create opportunities for analysing temporal patterns (T-patterns) and network structures. Positional data with high spatial and temporal resolution of different players at the same time open up to the analysis of interactions of players. A dynamical system approach seems a promising framework to study complex interaction between players in ball team sports. Analytical tools/methods from the dynamical system approach are ideal and pertinent because they can cope with this type of data. This approach may lead to new insights into the interactions of players within different ball team sports. Acknowledgements We would like to thank Richard Dik and Lisanne Rozema for their contribution in analysing the data presented in this chapter. References Araújo, D., Davids, K., Bennett, S. J., Button, C., & Chapman, G. (2004). Emergence of sport skills under constraints. In A.M.Williams & N. J. Hodges (Eds.), Skill Acquisition in Sport: Research, Theory and Practice (pp ). London: Routledge, Taylor & Francis. Barbero-Alvarez, J. C., Soto, V. M., Barbero-Alvarez, V., & Granda-Vera, J. (2008). Match analysis and heart rate of futsal players during competition. Journal of Sports Sciences, 26(1), Carling, C., Williams, A. M., & Reilly, T. (2005). Handbook of Soccer Match Analysis: A Systematic Approach to Improving Performance. London: Routledge. Carling, C., Reilly, T., & Williams, A. M. (2009). Performance Assessment for Field Sports. London: Routledge. Frencken, W. G. P. & Lemmink, K. A. P. M. (2008). Team kinematics of small-sided soccer games: a systematic approach. In T.Reilly & F. Korkusuz (Eds.), Science and Football VI (pp ). London and New York: Routledge. Haken, H., Kelso, J. A. S., & Bunz, H. (1985). A Theoretical-Model of Phase- Transitions in Human Hand Movements. Biological Cybernetics, 51,

16 Hughes, M., & Franks, I.M. (2008). The Essential of Performance Ananlysis: An Introduction. London: Routledge. Kelso, J. A. S. (1984). Phase-Transitions and Critical-Behavior in Human Bimanual Coordination. American Journal of Physiology, 246, Kugler, P. N., Kelso, J. A. S., & Turvey, M. T. (1980). On the concept of coordinated structures as dissipative structures. I : Theoretical lines of convergence. In G.Stelmach & J. Requin (Eds.), Tutorials in motor behavior (pp. 3-45). Amsterdam: North Holland. McGarry, T., Anderson, D. I., Wallace, S. A., Hughes, M. D., & Franks, I. M. (2002). Sport competition as a dynamical self-organizing system. Journal of Sports Sciences, 20, Passos, P., Araujo, D., Davids, K., Gouveia, L., Milho, J., & Serpa, S. (2008). Information-governing dynamics of attacker-defender interactions in youth rugby union. Journal of Sports Sciences, 26, Passos, P., Araujo, D., Davids, K., Gouveia, L., & Serpa, S. (2006). Interpersonal dynamics in sport: The role of artificial neural networks and 3-D analysis. Behavior Research Methods, 38, Schmidt, R. C., Carello, C., & Turvey, M. T. (1990). Phase-Transitions and Critical Fluctuations in the Visual Coordination of Rhythmic Movements Between People. Journal of Experimental Psychology-Human Perception and Performance, 16,