The
2012 World Congress on Performance Analysis in Sport Will G Hopkins1, Rita
M Malcata1,2, Sian V Allen1,2, Kirsten N Spencer1
Sportscience 16, 32-37, 2012 (sportsci.org/2012/WCPAS.htm) 1 Sport Performance Research Institute New Zealand, Millennium Campus, AUT University, Auckland, NZ; Email. 2 High Performance Sport New Zealand, Millennium Campus, AUT University, Auckland, NZ. Reviewers: Peter O'Donoghue, Cardiff School of Sport, Cardiff Metropolitan University, Cardiff, UK; Nic James, London Sport Institute, Middlesex University, London, UK.
|
|
The archetypal English country township of Worcester was the venue for this biennial conference of the International Society of Performance Analysis of Sport (ISPAS). The day before the conference started, three of us had the good fortune to wander into Worcester cathedral in time for choral evensong–a moving experience made even more poignant by the almost complete absence of a congregation. The zeitgeist moves on, the cathedral of sport is where you now find many worshippers, and some of us make our lives meaningful by analyzing sport performance. Derek Peters and his team from the University of Worcester provided a conference venue and logistics that were exceptional value for money, while two key members of ISPAS, Peter O'Donoghue and Nic James, organized and ran the scientific show with aplomb. We look forward to the next conference in 2014, wherever it might be. Any sport scientist interested in performance analysis (aren't we all?) should join the society to keep up with developments. Visit the ISPAS website for more. In the summaries below, we have included presentations only if there was a measure of performance or other data or insights relevant to the science or practice of performance analysis. We have used the word movements to mean distances covered or speeds of athletes that can be quantified from videos, accelerometers or global-positioning satellite (GPS) devices, while actions and plays refer to events or strategies that require coding of a live or recorded video data stream by an analyst. Alas, the keynote presentations did not have abstracts, so we have summarized only those we could attend. We have also provided paragraph summaries of a few noteworthy methodological developments, but otherwise only the sport and the relevant aspect of performance are listed under relevant headings–you will have to read the abstracts to get the findings. Download the book of abstracts and use the advanced search form in Acrobat (Ctrl-Shift-F) to find the abstract via the podium number (e.g., POD 1.1) or poster number (e.g., POST 1.1.1). The abstracts are sometimes in a foreign form of English that limits their utility. Some research presentations at the conference could be described as "games I have coded". Up to a point that's acceptable: this conference is partly a professional workshop providing performance analysts with a valuable opportunity to share and reflect on their experiences. But it was disappointing that the designs of the original-research studies were almost all descriptive: we could find only a handful of studies involving an experimental manipulation in which a performance measure was a moderator, mediator or outcome variable (POD 15.2, 16.4, 16.5; POST 1.1.5, 2.1.5), and all but one was a time series rather than more definitive crossovers or parallel-groups controlled trials. Maybe such high-quality research is happening but is too valuable to present at conferences–see the end of the Noteworthy Methodologies section. Keynotes
Accurate, meaningful content-rich data "anytime, anyplace, anywhere" (the Martini principle of marketing in the 1970s) via smartphones and other recent developments in information technology was the main message in Tony Kirkbride's enthralling keynote on media channels for performance analysis. Real-time or near real-time competition data of some high-profile sports events are available free, but high-quality data feeds require subscriptions with providers such as Opta. At the level of your own athlete or team, real-time interactivity between analyst, coach and athlete is also becoming a reality in what Tony called Martini-II. Barry Drust and Andy Scoulding provided insights into performance analysis in two top soccer teams of the English premier league. The role of performance analyst has evolved from that of simple match reporter to an indispensable source of information on monitoring of training, pre- and post-match analyses, and scouting for new players. The purpose of training monitoring is load management, providing useful information for coach and strength and conditioning staff. Pre- and post-match analyses rely on the quality of information and delivery. Piero, OPTA, Prozone, Amisco are some of the analysis tools in current use. Scouting for players has been made easier with players' profiling. Current challenges (and opportunities for upcoming analysts): how to extract relevant information from the deluge of data now available, and how to integrate it into day-to-day practice. Brian Dawson's keynote on movement patterns in team sports drew on his experience with Australian-rules football (AFL). The data from GPS devices have established relationships between training/game loads and injury risk and have helped make training sessions more specific to demands of the game, but relationships between measures of movement and competition performance are at best poor. Wynford Leyshon delivered a coach perspective for his keynote on the use of performance analysis in international 400-m hurdles. He used a case study of a London Olympian to demonstrate how analysts can best present race analysis data in order to help and engage coaches. Parameters of most interest to him included touchdown times and differences in pacing due to changes in stride patterns, which he referred to as "differential at change-down". Reports of automated tracking and analysis are always interesting, because hand coding of videos consumes huge amounts of time and only rudimentary analyses can be done for real-time or near real-time feedback to coach and players. Recent improvements were presented in a system that tracks players with data from two overhead cameras, the Sport Performance Analyzer (POD 6.1, 6.3, 6.6). Plays in basketball were classified correctly only ~80% of the time, and only ~90% with post-game manual processing. This system is promising but still some way from routine implementation. Monetary constraints often influence the level of performance analysis that is available to a team. An inexpensive option may be an inventive real-time notational analysis system where analysts use barcode scanning to code up to 15,000 different game events (POST 1.4.1). In initial trials analyzing soccer matches, scanners failed to read 0.9% of barcodes. An iPad application to configure, code and analyze live match performance also looks encouraging for the budget-conscious performance analyst, but the presenter failed to attend (POD 12.4). A "luck- and error-free" measure of team-sport performance has been derived by assigning a score to every action on the basis of what happened before and after the action (POD 3.6). The approach has been commercialized. When you have all those so-called performance indicators for your team, do they really tell you anything about team performance? According to Ed Burt and Mike Hughes, "numerous studies have assessed performance using action variables but have largely failed to consistently identify a link between individual performance indicators and match outcome." So they tried adding up counts and scores of various actions with weightings they assigned on the basis of their own understanding of the importance of the actions in the game of rugby union. Reassuringly, performance scores in 26 national-level matches of a national-league team and in 12 Six Nations matches showed "a very strong correlation with match outcome" (POST 1.2.10). We presume the scoring system could be adapted easily to the rating of individual players. It's also very likely that linear or neural-net models derived with large data sets will be even more successful. It's even more likely that such scoring systems have already been worked out in rugby and other sports and that you will never see them in conference abstracts or a journal! See for example this article about the role of analyst Ken Quarrie in the All Blacks' victory in the 2011 Rugby World Cup. Performance Indicators
POD 1.3 Performance rates (sum of good minus bad actions) of 287 players in the male World Championships. POD 1.4 Little difference in space-creation dynamics between age groups in 46 semi-final matches of a state championship. POD 3.5 Defensive plays in 80 matches of a national league. POD 6.1, 6.3, 6.6 Automated tracking of players with two overhead cameras in 4 games of a national professional A league. POD 12.4 An iPad application for live match analysis (presenter absent). POST 2.2.12 Defense-offense transitions in nine regional under-14 matches. Other
POD 1.1 Critical episodes in 80 matches from a national league. POD 1.2 Entropy as a measure of uncertainty in game scores in "18 NBA and 14 ACB regular seasons". POST 2.2.11 Time between scores in five NBA seasons. Combat Sports
Performance Indicators
POD 11.2 Actions and scoring with introduction of electronic trunk protector in n=? taekwondo international bouts. POD 11.3 System for real-time coding of judo on palm-tops and smart-phones established n=? bouts. POD 11.5 Combinations of punches in 8 bouts from each of 3 weight divisions of boxing. POD 11.6 Movements of elite female foil fencers in 100 international bouts. Other
POD 11.1 Errors in judging in 10 Olympic boxing bouts. Performance Indicators
POD 16.2 Body build and fitness of 16 female riders at three competitive levels. POD 16.3 Split times and performance in 27 successful clearance rounds and 49 with faults in a national (?) championship. Other
POD 16.1 "Large scale" survey of motivation in under-18 riders. POD 16.4 Effect of two saddle designs on posture in a crossover with nine experienced riders. POD 16.5 Effect of Pilates training on posture in an uncontrolled study of 10 riders. Performance Indicators
POD 8.1 Descriptive stats from 344 team timeouts in a national league. POD 8.3 Game statistics of winners and losers in five European men’s handball championships. POD 8.5, POST 1.2.9 Movements and actions of 69 players in three women's qualifying games of the European championships. POST 2.2.5 Actions and plays in six games of Valencian handball. Other
POD 8.2 Accuracy of referees (10) and their examiners (3). POD 8.4 Fitness and skills of 108 players at college level. POD 8.6 Movements of referees in the nine finals of a national league. Performance Indicators
POD 7.1 First-serve success and other indicators in 1929 Grand Slam tennis matches. POD 7.2 Duration of groundstroke as predictor
of aggressive or defensive shot in 567 rallies in women's tennis. POD 7.4 Gender and affect (and
implications for coaching) for eight mixed-pairs county badminton players. POD 7.5 Strokes and footwork in 10 male and five female top-30 table tennis matches. POD 14.3 Movements in two top-level matches of Padel
tennis. POST 2.3.4 Posture
in 2470 ground
strokes in four matches of men's Masters Cup tennis. POST 2.3.5 Set-victories momentum in three years of Grand Slam tennis. POST 2.3.7 Serve outcomes in a Grand Slam vs national junior tennis tournament. Other
POD 7.3 Physiological intensity of shot types of 12 recreational male squash players. POST
2.3.6 Movements
in one elite tennis
match. POST 2.3.9 Changes in rally length and score with
rule changes between a 2003 and 2010 squash championship. POST 2.3.8, POD 6.2 Analysis system for real tennis (the peculiar progenitor of tennis). Performance Indicators
POD 6.4 GPS for position-dependent movements in 98 players from eight premier clubs. POD 15.2 GPS for movements of 40 experienced vs novice players in small-sided games. POD 15.3 Network analysis for role of two-man tackles in producing turnovers in 20 matches of the World Cup. POD 15.4 Tackling in matches of a Super 15 rugby season. POD 15.5 Positional profiles in 28 home vs away games in the English championship. POD 15.6 Rucks in 15 Six-Nations matches. POST 1.2.10 Scores for actions predicting outcome in 26 national-level matches and in 12 Six-Nations matches. Other
POD 15.1 Kinematics of n=? kicks of 5 expert rugby and 5 expert Gaelic football players. POST 1.2.11 Clean and unclean scrums in 46 World-cup matches. POST 1.2.12 Tries scored per unit of possession time (or its inverse?) in 15 Six-Nations games. POST 1.2.12 Scoring and actions in the top and bottom nations of the World Cup. Performance Indicators–Player
POD 4.4 Effects of temperature, playing surface, travel, and recovery time on performance indicators in four years of the US major league. POD 5.1, POD 5.5, POST 2.1.1 Home advantage in Brazilian, Norwegian and English national leagues. POD 5.2 Normative profiling of young footballers in 30 matches. POD 5.3 Players’ spatial-temporal relationships in two national matches. POD 5.4, POD 9.1, POD 13.1 Offensive plays of two international teams in 24 national games. POD 9.2 Placing of the corner kick and outcome in 19 under-17 matches. POD 9.3, POD 13.4, POST 1.1.4 Movements and plays in small-sided games at various levels. POD 9.6 Attacking actions and plays in 60 matches of three national leagues. POD 13.5 Analysis of goals in 41 matches of a youth academy team. POD 12.6 Differences in movements of five semi-elite players between GPS, manual tracking, and video-based analysis systems. POST 1.1.1 Effect of relative numbers of players in attackers' and defenders’ plays in 10 international games and n=? lower-level practice games of futsal. POST 1.1.3 Poor relationships between 22 action variables and outcomes in 277 national championship matches. POST 1.1.5 Effect of zone vs man-to-man defenses on plays in a trial of 12 under-17 elite players in small-sided games. POST 1.4.1 Barcode scanning for notational analysis validated with soccer matches. POST 2.1.2 Normative profiling of 136 corner kicks in 12 matches in a professional league. POST 2.1.4 Time and pitch zone of 439 goal scoring opportunities in semi-elite women’s games. POST 2.1.5 Effect of limiting ball touches on spatial interaction in a trial of 18 under-17 elite players in small-sided games. POST 2.1.8 Network analysis to identify attacking plays in two teams of a premier league. Performance Indicators–Goal Keeper
POD 2.5 Analysis system for the goal keeper developed from 15 interviews. POD 9.4 Comparison between training (1 wk) and match actions (92 games) of goal keepers. POST 2.1.3 Network analysis for role of goalkeeper in 61 passing sequences in three matches of premier league. Coach
POD 2.2 Behavior of three head coaches toward 48 effective and non-effective players in 72 training sessions. POD 2.3 Observation effectiveness and knowledge of players and the game in interviews of eight national-league coaches. Other
POD 2.4 Perceived role of performance analysis from interviews of 8 players and three managers in one professional club. POD 4.5 Effect of environmental temperature on home advantage in 346 international matches. POD 9.5 Release of players to World Cup and ensuing clubs' performance. POST 2.1.11 Career achievements of 154 international under-17 players. POST 2.1.12 Quantification of injuries and identification of high-risk plays in 160 youth football matches. Volleyball
Performance Indicators
POD 3.4 Successful actions of the setter in 24 championship games. POD 10.5 Defensive actions in five national league games. POST 1.2.2 Determinants of attack tempo in 19 high-level male games. POST 1.2.3 Winning serves in n=? championship games. POST 1.2.4 Successful areas of attack in 10 games of men’s professional beach volleyball. POST 2.2.1, POST 2.2.4 Actions of winning teams in 31 sets of female beach volleyball. Other
POST 1.2.5 Retrospective survey of training of 229 female and male adult players during earlier stages of development. Other Individual Sports
Performance Indicators
POD 14.2 1RM strength and shot-put performance of 29 females and 24 males at national collegiate level. POD 14.4 Movement elements, synchronicity and space utilization of rhythmic gymnasts in three international championships. POST 2.3.2 Pivots and other movements of eight rhythmic gymnasts in a national final. POD 14.6 Pack size and finishing position in 305 world-championship triathlon races. POD 11.4 Body sway and shooting accuracy in 117 biathletes. of various ages. POST 1.4.2 Effects of environment (wind and waves) on boat speed in an international rowing regatta (presenter absent). POST 1.4.3 Predicting javelin performance from 438 throws of 20 national-level athletes using neural network modeling. POST 1.4.10 Subjective marking vs objective computerized scoring of 13 national-level trampolinists. Other
POD 2.1 Case study of experience of a golf coach as a performance analyst. POD 14.1 "ISOPAR" method to estimate difficulty of any position
on a golf
course. POD 14.5 "Help intensity" and incentives in Tour de France cyclists since 1947. POST 1.4.8 A kinematic analysis of the Yurchenko vault in 18 female gymnasts. POST 2.3.3 Needs analysis from n=? interviews of coaches and athletes in canoe and kayak slalom. POST 2.5.8 EMG activity during acrobatic back handspring performance in five gymnasts. Performance Indicators
POD 3.2 Using small-sided games to increase exposure to actions and plays for development of young netballers. POD 3.3 Analysis of movements and performance in Australian-rules football. POST 1.2.7 Accuracy of decision-making in 13 games of Australian-rules football. POD 10.3 Analysis of kicking and striking skills in Gaelic football. POST 1.2.6 Sixty kinds of actions of 29 winning and 30 losing teams in male (?) domestic 20-20 cricket. POST 1.2.8 Movement intensities via heart rates in 15 ice-hockey players. POST 2.2.6 Actions in three under-18 national-level ice-hockey matches. POST 2.2.9 Players' playing time at the water polo World Championships. POST 2.2.10 Momentum in shooting during 20 national netball matches. Other
POD 2.6 Time constraints and efficacy of performance analysis in n=? coaches and teams of National v International netball. POD 3.1 Development of a test simulating an innings of cricket batting. POD 10.1 Accuracy of umpires’ decisions in netball. POD 4.3 A biomechanical model for movements causing ACL injury. POD 4.1 Monitoring fatigue in 11 trained biathletes with an orthostatic heart-rate test. POD 10.6 Retrospective survey of sporting and academic performance of 242 athletes of mainstream and elite sport schools. Acknowledgements: Teesside University, High Performance Sport NZ and
Will's consulting account all contributed to some expenses for this
conference tour, while AUT University provided salary for Will and Kirsten. Published October 2012 |