Mixed Martial Arts Bout Prediction Using Artificial Intelligence

Abstract

AbstractIn modern times, boxing is not the only sport that is based on martial arts. The competitor to boxing has recently risen in the form of mixed martial arts (MMA), which is becoming quite a famous sport, most notably recognized due to the Ultimate Fighting Championship promotion (UFC). As with most other competitive sports, it is never obvious who will emerge as the winner of a specific MMA bout, as the sport is very unpredictable by nature. This paper focuses on using machine learning (ML) algorithms and classifiers to predict the winners of MMA bouts in the Ultimate Fighting Championship promotion. The data used to train and test the algorithms and classifiers is sampled from the rich UFC database which includes details from all bouts that have taken place in the UFC throughout the history of the promotion. The predictions are obtained without using any data that can directly indicate the winner of a bout. The results are presented through accuracy percentages for two different types of sample sets. Successful bout prediction percentages range from 80% to just over 92% across all subsets. This model can be further expanded and modified to predict the winner for future matchups in the promotion.KeywordsMachine learning (ML)Ultimate Fighting Championship (UFC)Mixed martial arts (MMA)Bout prediction