Rhythms of Victory: Predicting Professional Tennis Matches Using Machine Learning

Abstract

Forecasting the winning matches of professional tennis players has a wide range of practical applications. We introduced an innovative approach to quantify and combine strategic and psychological momentum using the entropy weight method and analytic hierarchy process, and tested its effectiveness. Utilizing data from the Wimbledon Championship 2023, we constructed a support vector machine model to predict the turning point and winner of each point, and optimized it using particle swarm optimization. Our model achieved a significant level of accuracy (96.09% for turning point and 83.52% for predicting the winner) and performed well in different courts and players. Furthermore, we compared its performance with commonly utilized predictive models, including ARIMA, LSTM and BP networks, and found that our model exhibited higher accuracy than other existing models in predicting the point winner. Our study provides a reference for the role of momentum in dynamic matches, and our model can be used to calculate the odds of tennis matches and provide guidance to coaches.

Keywords

• Forecasting
• support vector machine
• machine learning
• particle swarm optimization
• tennis