Momentum Prediction for Tennis Matches Based on Counter-Factual Analysis and Multi-LGBM

This research paper investigates the concept of “momentum” in tennis and its impact on player performance. To calculate “momentum,” grounded in counter-factual analysis, we develop a Multi-LGBM (Light Gradient Boosting Machine) chain model, which incorporates the calculations of multi-probability and “leverage.” Our analysis, based on continuous scoring and win probabilities, uncovers a strong correlation between momentum and player performance, highlighting its critical role in the dynamics of tennis matches.