Introduction In the last decade, the International Basketball Federation has focused on popularizing 3x3 basketball worldwide. Despite these efforts, there is still limited knowledge of notational analysis in the sport. Due to the dynamic nature of 3x3 basketball, univariate analysis is insufficient to understand team performance. This research aims to identify performance indicators and contextual variables that impact game outcomes in professional 3x3 basketball. Methods Data from 3x3 basketball World Tour games (2015-2022) featured 13 normalized performance indicators and 3 contextual variables (Memmert, 2024). Logistic Regression, Decision Tree, and Multilayer Perceptron Neural Network models classified winning and losing teams, with performance assessed by the area under the curve. Results The neural network outperformed the other two models, with over 80% of accuracy (AUC=0.89). The effect of the opponent quality and four key performance indicators (KPIs), incorporating the percentage of both 2-points and 1-points (calculated by dividing shots made by shots attempts respectively), defensive rebounds, and turnovers, on game outcome was detected in all situations. Additionally, the performance indicators, including team fouls, extra free throws, key assists, and ball possession affect game outcomes. Discussion While four KPIs were identified, the winners-losers paradigm may oversimplify the game, missing nuanced player and strategy contributions. The percentage of 1-points and 2-points were identified as KPIs. Unlike 5x5, 3x3 players took more 6.75m shots, likely due to the double benefit of long shots in 3x3 versus 1.5 times in 5x5. Defensive rebounds, turnovers, and contextual variables like opponent quality influence the outcomes, similar to 5x5 basketball (Csataljay et al., 2020). Due to the specific foul rule, emerging metrics like extra free throws and key assists contribute to winning the game. Differences were observed between winning and losing teams in all performance indicators except for driving. Limitation This study failed to include female games, which limited the comprehensive understanding of 3x3 basketball outcomes due to the anthropometrical differences.
© Copyright 2024 15. Symposium der dvs-Sektion "Sportinformatik und Sporttechnologie": Zwischen Geistesakrobatik und praktischer Anwendung: Innovationen in der Sportinformatik und Sporttechnologie - Abstractband. Veröffentlicht von Technische Universität Dortmund. Alle Rechte vorbehalten.