Research on Prediction of Key Points in Tennis Matches Based on Neural Network and Momentum Analysis

Abstract

In modern sports, advanced data analysis sheds light on match dynamics. Our study focuses on momentum and leverage in Wimbledon 2023 matches using ATennision Model, a neural network model with Transformer architecture. We analyze winning transitions and key turning points by tracking real-time changes in winning percentages, our model obtains an average accuracy of 94.1%. Leverage, the impact of specific points on overall match outcomes, is a key metric. We quantify leverage effects and assess momentum's influence on match flow. Statistical tests challenge the belief that momentum is insignificant. We develop a Turning Point Prediction model for strategic guidance and achieve high predictive accuracy. Our study enhances understanding of tennis dynamics and equips players with strategies to capitalize on or counteract momentum shifts, aiming to decode winning transitions in this dynamic sport.