Multidimensional technical performance in the 2024 Paris Olympic Women’s Basketball Tournament: an entropy−weighted rank−sum ratio evaluation with a USA–France final case analysis

Abstract

Elite women’s basketball performance is multidimensional, yet transparent tournament−level approaches for integrating heterogeneous technical indicators remain limited. This study applied an entropy−weighted rank−sum ratio (WRSR) framework to the 2024 Paris Olympic Women’s Basketball Tournament and used the same 11−indicator framework in an embedded descriptive analysis of the United States–France final. Official International Basketball Federation (FIBA) box scores from all 26 games yielded 52 team−game records for data auditing and indicator construction. Structural screening reduced 21 eligible candidates to 11 indicators, all of which were retained following descriptive correlation reviews at the team−game and 12−team model levels. The 12 team−level tournament profiles formed the WRSR evaluation matrix. Equal−weight and leave−one−indicator−out specifications assessed sensitivity to analytical choices, and Spearman rank correlation assessed same−tournament correspondence with the official final classification. Each finalist’s final−game values were also compared with its own five−game pre−final profile. No indicator pair reached the near−redundancy review flag of |r| ≥ 0.90 at either analytical level; the largest model−level correlation was between points per game and two−point percentage ( r = 0.895). The largest entropy weights were assigned to assists (0.1396), personal fouls (0.1265), and two−point percentage (0.1141). The United States ranked first by WRSR (0.8895), followed by Australia (0.6715) and Belgium (0.6452). The equal−weight ranking was closely aligned with the main ranking (ρ = 0.963, p < 0.001). Leave−one−indicator−out correlations ranged from 0.923 to 1.000, with some local rank and top−three changes. The WRSR ranking showed positive but incomplete correspondence with the official classification (ρ = 0.769, p = 0.003). In the final, points, three−point percentage, and assists were below the respective pre−final ranges for both teams. The entropy−weighted WRSR framework produced a transparent, context−specific ordering of multidimensional technical performance with high overall rank consistency. Selected indicator omissions nevertheless produced local changes. The official−ranking comparison provided a same−tournament consistency assessment, and the embedded final analysis described within−team departures from preceding performances. Replication across competitions and incorporation of possession−, opponent−, and game−context adjustments are needed to evaluate generalizability.