Sullivan, C, Kempton, T, Ward, P & Coutts, AJ 2020, 'The efficacy of talent selection criteria in the Australian Football League.', Journal of sports sciences, vol. 38, no. 7, pp. 773-779.View/Download from: Publisher's site
This study investigated the association between talent selection criteria, draft order and match performance in professional Australian Football players. Physical performance results from the Australian Football League (AFL) National Draft combine and non-performance based talent selection criteria were collated for all players drafted in the National Draft with selections 1-80 between 2003 and 2008 (n = 318). Match performance was assessed via the AFL Player Ranking metric that was provided by a commercial statistical provider (Champion Data Pty Ltd). A combination of stepwise multiple regression and linear mixed model analyses examined the influence of National Draft combine physical performance assessments and non-performance based talent selection criteria on draft order and future match performance. Earlier draft selection was associated with Under-18 all Australian team selection, height, Under-18 National Championship participation and indigenous status. The 20 m sprint and state of origin were associated with later draft selection. Under-18 all Australian team selection increased both Player Ranking/game and total Player Ranking. Under-18 all Australian team selection displays efficacy for selecting players with the potential for success.
Ward, P, Windt, J & Kempton, T 2019, 'Business Intelligence: How Sport Scientists Can Support Organization Decision Making in Professional Sport.', International journal of sports physiology and performance, vol. 14, no. 4, pp. 544-546.View/Download from: Publisher's site
The application of scientific principles to inform practice has become increasingly common in professional sports, with increasing numbers of sport scientists operating in this area. The authors believe that in addition to domain-specific expertise, effective sport scientists working in professional sport should be able to develop systematic analysis frameworks to enhance performance in their organization. Although statistical analysis is critical to this process, it depends on proper data collection, integration, and storage. The purpose of this commentary is to discuss the opportunity for sport-science professionals to contribute beyond their domain-specific expertise and apply these principles in a business-intelligence function to support decision makers across the organization. The decision-support model aims to improve both the efficiency and the effectiveness of decisions and comprises 3 areas: data collection and organization, analytic models to drive insight, and interface and communication of information. In addition to developing frameworks for managing data systems, the authors suggest that sport scientists' grounding in scientific thinking and statistics positions them to assist in the development of robust decision-making processes across the organization. Furthermore, sport scientists can audit the outcomes of decisions made by the organization. By tracking outcomes, a feedback loop can be established to identify the types of decisions that are being made well and the situations where poor decisions persist. The authors have proposed that sport scientists can contribute to the broader success of professional sporting organizations by promoting decision-support services that incorporate data collection, analysis, and communication.
Sullivan, C, Kempton, T, Ward, P & Coutts, AJ 2018, 'Factors associated with early career progression in professional Australian Football players.', Journal of sports sciences, vol. 36, no. 19, pp. 2196-2201.View/Download from: Publisher's site
This study examined the association between individual and team characteristics and the probability of being offered a second contract in professional Australian Football. Contract status was obtained from the AFL for players who were drafted in the AFL National Draft between 1999 and 2013 (n = 999). Individual player characteristics were retrieved from the AFL while variables relating to performance were accessed online via Champion Data®. A binary logistic regression examined the influence of each characteristic on the probability of a professional Australian Football player receiving a second contract. Receiver operating characteristic (ROC) curves and the associated AUC were used to assess the discriminant ability of both a training (n = 938) and test data set (n = 61). The characteristics that influenced the probability of receiving a second contract included first year debut (pr 0.606), draft order (pr - 0.126), draft year (pr 0.059), games played (pr 1.848), team state (pr 0.458), rising star nomination (pr 1.553) and team ladder position (pr -0.043) (χ2 (8) = 198.28, p < 0.001). The ROC curve demonstrated an AUC of 82.4% (training) and 76.0% (test). A combination of individual and team based characteristics are associated with early career progression in professional Australian Football.
Ward, P, Coutts, AJ, Pruna, R & McCall, A 2018, 'Putting the "I" Back in Team', INTERNATIONAL JOURNAL OF SPORTS PHYSIOLOGY AND PERFORMANCE, vol. 13, no. 8, pp. 1107-1111.View/Download from: Publisher's site
Ward, PA, Ramsden, S, Coutts, AJ, Hulton, AT & Drust, B 2018, 'Positional Differences in Running and Non-Running Activities During Elite American Football Training', Journal of Strength and Conditioning Research, vol. 32, no. 7, pp. 2072-2084.View/Download from: Publisher's site
The aim of this investigation was to describe differences in training loads between position groups within professional American football. Integrated micro technology data was collected on 63 NFL football players during an American football training camp. Five key metrics (total distance, high speed distance, Player Load, Player Load per Minute, and Total Inertial Movement Analysis (IMA)) served to quantify both running and non-running activities. Players were classified into position groups (DB, DL, LB, OL, QB, RB, TE, and WR). Training sessions were identified by their relationship to the upcoming match (e.g., -4, -3, -2). Running and non-running activities varied between position groups relative to the training day. Differences in total distance were between DB and WR were observed to be unclear between the three training days (Game Day (GD) -4: 74 ± 392 m; GD -3: -122 ± 348; GD - 2: -222 ± 371 m). However, moderate to large differences were observed between these two positions and the other positional groups. A similar relationship was observed in Player Load and Player Load per Minute, with the DB and WR groups performing greater amounts of load compared to other positional groups. Differences in High Speed Distance varied across positional groups, indicating different outputs based on ergonomic demands. The OL and DL groups ran less but engaged in a higher amount of non-running activities (Total IMA) with differences ranging from moderate to large across the three training days. Total IMA differences between offensive and defensive linemen were unclear on GD -4 (-4 ± 9) and GD -2 (-2 ± 8) and likely moderate on GD -3 (-9 ± 9). Positional differences with regard to running and non-running activities highlight the existence of position specific training within a training micro-cycle. Additionally, Total IMA provides a useful metric for quantifying sport specific movements within the game of American football.