The Tennis Momentum Prediction and Analysis Based on The LightGBM Model

Tennis is one of the most popular sports worldwide. Audiences love tennis matches that are full of excitement and swings. In the match, to capture the flow of play is of great importance to both the coaches and the athletes themselves. In this paper, we analyze the dataset of Wimbledon 2023 Gentlemen’s matches and develop a model related to “momentum” with LightGBM, which captures the flow of play. We first introduce some indexes, such as net point-won rate, and the number of breakpoints, which are related to “momentum”. The proposed model is constructed using LightGBM, and the dataset is utilized to train and validate the proposed model. Furthermore, some common metrics, such as AUC, TPR, and FPR, are used to prove the superiority of our model. AUC is the area under the ROC curve, the proportion correctly predicted to be positive and the proportion incorrectly predicted to be positive. Finally, we apply the proposed model to other men's tennis matches, and it also has a good prediction effect on other matches, which verifies that our model has strong generalization performance.