Research on Momentum Evaluation and Prediction Based on LightGBM and SVC Models

This paper primarily investigates the application of the momentum concept in the field of sports, focusing particularly on its utilization in tennis matches. We conducted an in-depth analysis using match data from the 2023 Wimbledon Championships and developed a predictive model to capture momentum variations during matches and evaluate its impact on match outcomes. Initially, we established a real-time player performance assessment system based on the LightGBM model, calculating the importance of each feature to demonstrate the correlation between momentum and actual player scores. Subsequently, after comparing common machine learning models, we selected the SVC model and assessed the importance of features in the SVC model using the information gain algorithm, further analyzing the influence of momentum on match outcomes. Finally, we predicted the outcomes of four random matches and discussed the stability of the model. This paper offers a new perspective on the role of momentum in tennis matches and proposes scientific training and match preparation approaches, providing robust reference points for athletes and coaches to optimize match strategies.