Development and validation of a player impact ranking system in Australian football

Abstract

This study aimed to develop and validate player performance impact rankings for Australian football, considering players’ time on ground and game situation. Player performance data was collected from an Australian Football League (AFL) club and their opponents in each game during the 2006 season. Individual player and team impact scores were generated by multiplying the frequency of selected game actions by allocated positive or negative numerical values. The study was divided into three phases. In phase 1 higher team impact scores were shown to have a significant correlation with winning (r=−0.69, p<0.01). A greater final points margin between the teams was also correlated with an increased impact score margin (r=0.85, p<0.001). In phase 2 one-way ANOVAs revealed individual player impact scores were significantly higher in the midfield than in the forward and defensive positional zones (p<0.001), suggesting that impact score comparisons should only be made within positional zones. In phase 3 a chi-square analysis revealed significant differences between individual players within each of the positional zones. It was concluded that the impact ranking scores provided a valid method of assessing game performance for players (within positional zones) and teams, allowing performance profiles to be created for coaching purposes.