RESUMO
âIn team sport, classifying playing position based on a players' expressed skill sets can provide a guide to talent identification by enabling the recognition of performance attributes relative to playing position. Here, elite junior Australian football players were a priori classified into 1 of 4 common playing positions; forward, midfield, defence, and ruck. Three analysis approaches were used to assess the extent to which 12 in-game skill performance indicators could classify playing position. These were a linear discriminant analysis (LDA), random forest, and a PART decision list. The LDA produced classification accuracy of 56.8%, with class errors ranging from 19.6% (midfielders) to 75.0% (ruck). The random forest model performed at a slightly worse level (51.62%), with class errors ranging from 27.8% (midfielders) to 100% (ruck). The decision list revealed 6 rules capable of classifying playing position at accuracy of 70.1%, with class errors ranging from 14.4% (midfielders) to 100% (ruck). Although the PART decision list produced the greatest relative classification accuracy, the technical skill indicators reported were generally unable to accurately classify players according to their position using the 3 analysis approaches. This player homogeneity may complicate recruitment by constraining talent recruiter's ability to objectively recognise distinctive positional attributes.
Assuntos
Desempenho Atlético/classificação , Destreza Motora/classificação , Futebol/classificação , Adolescente , Aptidão , Austrália , Humanos , Análise e Desempenho de TarefasRESUMO
This study investigated the evolution of game-play manifested via team performance indicator characteristics in the Australian Football League (AFL) from the 2001 to 2015 seasons. The mean values for 18 performance indicators were collated for every AFL team over 15-seasons. A multivariate analysis was used to uncover temporal trends in the dataset. Compared to the 2004 season, the 2005 to 2010 seasons were characterised by large growth in the counts of handballs (d = 0.83; 90% CI = 0.22-1.43), disposals (d = 1.24; 90% CI = 0.59-1.87), uncontested possessions (d = 1.37; 90% CI = 0.71-2.01), clangers (d = 2.14; 90% CI = 1.39-2.86) and marks (d = 1.43; 90% CI = 0.76-2.07). Contrastingly, the effective disposal percentage declined rapidly during the same period. The number of inside 50 m counts remained stable throughout the 15-season period. The ordination plot of league-wide performance indicator characteristics illustrated a distinct cluster from the 2001 to 2004 seasons, an abrupt shift from the 2005 to 2009 seasons, and an emergent (re)stabilisation from the 2010 to 2015 seasons. The results demonstrate the synchronous league-wide evolution of game-play in the AFL from the 2001 to 2015 seasons. Amongst other constituents, this evolution likely reflects the introduction of modernised coaching strategies, rule changes and changing perceptions of rule interpretations.
Assuntos
Desempenho Atlético/tendências , Futebol Americano/tendências , Austrália , HumanosRESUMO
Physical testing-based draft combines are undertaken across various sporting codes to inform talent selection. To determine the explanatory power of the Australian football league (AFL) draft combine, participants drafted between 1999-2016 (n = 1488) were assessed. Testing performance, draft selection order and playing position, AFL matches played, AFL player ranking points and AFL player rating points were collected as career outcomes. Boosted regression tree analysis revealed that position and draft selection order were the most explanatory variables of career outcomes. Linear modelling based on testing results is able to explain 4% of matches played and 3% of in-game performance measures. Each individual combine test explained <2% of the matches played outcome. Draft selection order demonstrated mixed results for career outcomes relative to playing position. For instance, key forwards and draft selection order were observed as a slight negative relationship using the AFL Player Ranking points career outcome measure. These findings indicate that the AFL draft combine is a poor measure for informing talent selection, thus providing minimal utility for the practices investigated in this study.
Assuntos
Desempenho Atlético , Esportes , Adolescente , Humanos , Masculino , Adulto Jovem , Aptidão , Desempenho Atlético/classificação , Desempenho Atlético/estatística & dados numéricos , Desempenho Atlético/tendências , Austrália , Teorema de Bayes , Estudos Transversais , Previsões/métodos , Modelos Lineares , Análise de Regressão , Estudos RetrospectivosRESUMO
OBJECTIVES: Analysing the dissimilarity of seasonal and team profiles within elite sport may reveal the evolutionary dynamics of game-play, while highlighting the similarity of individual team profiles. This study analysed seasonal and team dissimilarity within the National Rugby League (NRL) between the 2005 to 2016 seasons. DESIGN: Longitudinal. METHODS: Total seasonal values for 15 performance indicators were collected for every NRL team over the analysed period (n=190 observations). Non-metric multidimensional scaling was used to reveal seasonal and team dissimilarity. RESULTS: Compared to the 2005 to 2011 seasons, the 2012 to 2016 seasons were in a state of flux, with a relative dissimilarity in the positioning of team profiles on the ordination surface. There was an abrupt change in performance indicator characteristics following the 2012 season, with the 2014 season reflecting a large increase in the total count of 'all run metres' (d=1.21; 90% CI=0.56-1.83), 'kick return metres' (d=2.99; 90% CI=2.12-3.84) and decrease in 'missed tackles' (d=-2.43; 90% CI=-3.19 to -1.64) and 'tackle breaks' (d=-2.41; 90% CI=-3.17 to -1.62). Interpretation of team ordination plots showed that certain teams evolved in (dis)similar ways over the analysed period. CONCLUSIONS: It appears that NRL match-types evolved following the 2012 season and are in a current state of flux. The modification of coaching tactics and rule changes may have contributed to these observations. Coaches could use these results when designing prospective game strategies in the NRL.
Assuntos
Desempenho Atlético , Futebol Americano , Estações do Ano , Austrália , Humanos , Estudos Longitudinais , Análise MultivariadaRESUMO
BACKGROUND: Learning transfer is defined as an individual's capability to apply prior learnt perceptual, motor, or conceptual skills to a novel task or performance environment. In the sport sciences, learning transfers have been investigated from an athlete-specific perspective. However, sport scientists should also consider the benefits of cross-disciplinary learning to aid critical thinking and metacognitive skill gained through the interaction with similar quantitative scientific disciplines. OBJECTIVE: Using team sports performance analysis as an example, this study aimed to demonstrate the utility of a common analytical technique in ecology in the sports sciences, namely, nonmetric multidimensional scaling. METHODS: To achieve this aim, three novel research examples using this technique are presented, each of which enables the analysis and visualization of athlete (organism), team (aggregation of organisms), and competition (ecosystem) behaviors. RESULTS: The first example reveals the technical behaviors of Australian Football League Brownlow medalists from the 2001 to 2016 seasons. The second example delineates dissimilarity in higher and lower ranked National Rugby League teams within the 2016 season. Lastly, the third example shows the evolution of game play in the basketball tournaments between the 2004 and 2016 Olympic Games. CONCLUSIONS: In addition to the novel findings of each example, the collective results demonstrate that, by embracing cross-disciplinary learning and drawing upon an analytical technique common to ecology, novel solutions to pertinent research questions within sports performance analysis could be addressed in a practically meaningful way. Cross-disciplinary learning may subsequently assist sport scientists in the analysis and visualization of multivariate datasets.