A dual-model framework for Olympic sports selection: integrating TOPSIS evaluation and logistic regression prediction

The International Olympic Committee (IOC) grapples with optimizing its sports roster amid competing priorities like youth engagement, sustainability, and tradition. This study addresses subjective selection limitations by developing a dual-model framework integrating TOPSIS for multi-criteria evaluation and logistic regression for retention prediction. Leveraging rigorously processed Olympic datasets (1988–2028), the model achieved 89.7% accuracy in forecasting recent sports changes (e.g., karate removal) with strong discriminative power (AUC = 0.92). For the 2032 Olympics, billiards emerged as the optimal candidate (TOPSIS score = 0.6585), excelling in safety and sustainability, while American football showed moderate potential (score = 0.4766). Sensitivity analysis revealed Olympic participation rates as dominant predictors. Critically, worldwide popularity had negligible impact. This research provides the IOC with the first integrated tool enabling evidence-based sports selection, identifies actionable interventions for at-risk sports (e.g., format modernization for legacy sports), and establishes quantifiable thresholds for future decision-making.