Research on Tennis Match Momentum Analysis Based on Random Forest and Hidden Markov Model

This paper explores the evaluation of tennis players' performance and the dynamics of match momentum through comprehensive statistical analysis. Employing Spearman correlation and Fisher discriminant analyses, the study establishes a performance evaluation model, highlighting first serve percentage, second serve percentage, and unforced errors as key performance indicators. Further, the paper adopts the least squares method and the Hidden Markov Model to elucidate the impact of momentum in tennis matches, identifying serving, consecutive games won, and errors as pivotal factors. The Mann-Kendall mutation test is utilized to pinpoint momentum shifts within matches, with a subsequent Random Forest analysis revealing the significant influence of serving, double faults, and match streaks on match outcomes. Evaluating the developed relative momentum model across various tennis and table tennis matches demonstrates its efficacy in mirroring actual match fluctuations, offering insights for potential model enhancements.