Game changers: An objective assessment of players' contribution to team success in women's rugby league.

Abstract

This study introduces new performance metrics to address the lack of objective player evaluations in women's rugby league. Using data from six seasons (2018-2023) of the Women's National Rugby League (NRLW), five machine learning algorithms generated two key metrics: "Wins Created" for offensive performance and "Losses Created" for defensive performance. These were adjusted by a situational importance modifier based on player positions and combined into a final metric called "Net Wins Added". An Elo rating variant modified to suit a rugby league context was also created to provide a strength of opponent multiplier for player performance. The validity of these metrics against traditional objective and subjective performance measures in rugby league were evaluated. The metrics predicted seasonal team wins with a Root Mean Squared Error (RMSE) of 0.9 and Player of the Year top 10 leaderboard points with an RMSE of 8.2. The metrics displayed substantial agreement (Gwet AC1 = 0.82) when predicting experts' Team of the Year award recipients and substantial agreement (Gwet AC1 = 0.75) when predicting players' Team of the Year awards. Developing and validating these objective player performance metrics provide women's rugby league with a unique system to enhance talent evaluation and player recruitment.