Determining the key performance indicators on game outcomes in NBA based on quantile regression analysis

The purpose of this study was to (i) determine the effect of KPIs on game outcomes for NBA teams; and (ii) compare the result difference between multiple linear regression (MLR) and quantile regression (QR) analysis. Data were collected for a total of 2,322 games in the 2020–2022 NBA regular seasons and play-in games. QR and MLR were adopted to analyse the relationship between KPIs and game outcomes. The results revealed that game outcomes were associated with two-point percentage, three-point percentage, free-throw percentage, offensive rebounds, assists, defensive rebounds, turnovers, steals, fouls, and game pace, but not with blocks, opponent quality, and game location. Differences were reported between MLR and QR analysis in free-throw percentage and fouls. This study showed that the QR analysis is more sensitive to identifying the effect of KPIs on game outcomes than MLR. The insights provided by QR analysis can guide more effective team strategies and player selection, emphasising the critical need for sports analysts and practitioners to consider the appropriate statistical methods when evaluating performance indicators in basketball and other sports.