Tennis Ball Momentum Study Based on CUMSUM Detection Algorithm and Machine Learning Models

Successful tennis players need to be technically skilled and physically fit. However, in tournaments, results are often unexpected, and this fluctuation is attributed to "momentum". The study of momentum factors can help players and coaches to better control the pace of the game and develop effective tactics to increase the chances of winning. In this study, we propose a method to calculate momentum based on the entropy weighting method, find turning points using the CUMSUM detection algorithm, verify the significant effect of momentum on the outcome of the game, and construct a prediction model using machine learning. We successfully apply it to real matches, especially in the 2023 Wimbledon men's final. Our model demonstrates high predictive accuracy through XGBoost integrated learning, providing reliable tactical advice for athletes and coaches.