Comparing the Sensitivity of Hierarchical Multiple Regression and Quantile Regression in Identifying Key Determinants of Winning in Elite Men's Basketball

This study systematically examined the effectiveness of stratified multiple regression (MLR) and quantile regression (QR) in identifying key performance indicators (KPIs) that influence the outcomes of elite men's basketball games. Performance data from the Paris 2024 Olympic Games was analysed by comparing MLR and QR across multiple quartiles to capture overall trends and distribution-specific nuances. Key predictors such as inside-out scoring, centre-back scoring, three-point scoring, free-throw scoring and pace of play were significant in both methods. However, QR reveals additional insights that MLR does not, such as the importance of Q50 Offensive Rebounds and Q75 Assists and Caps. QR demonstrates greater sensitivity in capturing the complex and context-dependent relationships between KPIs and performance outcomes, resulting in a nuanced understanding of how these factors differ across levels of competition. While MLR is effective in providing stable results, it has limitations in identifying this variation. This study highlights the critical role of quantitative research in advancing sport analysis, revealing subtle performance dynamics and providing practical guidance for optimising team strategies, tactical decisions and training design. By integrating advanced statistical methods, this study contributes to a deeper understanding of basketball success and establishes a methodological framework for future research in sports performance analysis.