Abstract
Basketball preparation has shifted from a physical and technical focus to a holistic approach that incorporates performance analysis, as traditional statistics offer only a limited understanding of team behavior. This study aimed to characterize NBA teams according to their performance in regular season and play-off games and to identify the play types that distinguish the best-performing teams in each phase. Data from five NBA seasons (2019–2024; 6400 games) were analyzed using play-type statistics obtained from the official league database. Two-step cluster analysis and one-way analyses of variance with Bonferroni correction were applied to identify group differences (p < 0.05). Three team clusters were identified in both the regular season and the play-offs. High-performing teams in the regular season were significantly more effective in isolation (p < 0.01) and spot-up (p = 0.03) situations and showed greater use of pick-and-roll ball-handler actions (p = 0.001). In the play-offs, differences were smaller and mainly involved low-performing teams, which were less effective in transition and spot-up plays (p < 0.05). Comparisons between the best regular season and play-off teams revealed significant differences in post-up, off-screen, and put-back efficiency (p < 0.05). Success depended primarily on execution efficiency rather than play-type frequency.