Optimizing Fighter Strategies and Predicting Outcomes in Bellator MMA Using Artificial Intelligence

Artificial intelligence (AI) has seen significant growth in sports applications, ranging from outcome prediction and tactical decision support to enhancing referee precision and preventing athlete injuries. This study investigates the use of machine learning algorithms to forecast the results of Bellator MMA matches and identify the key determinants affecting these outcomes. The dataset, comprising over 300 bouts held between 2010 and 2016, offers detailed fighter statistics ideal for machine learning analysis. We applied Random Forest (RF), Gradient Boosting Decision Trees (GBDT), and Support Vector Machines (SVM) to predict match results and highlight the top 10 influential features. Performance comparisons, based on metrics such as accuracy, Cohen's kappa, sensitivity, and specificity, indicated that RF and GBDT improved when using selected key features, whereas SVM showed reduced accuracy and stability. Despite variations in feature importance across models, fight date and age differences between fighters consistently ranked among the top three.