Author(s)

Guowei Zhou, Yanxi Fan

Corresponding Author

Guowei Zhou

Abstract

Tennis matches involve complex interactions between players, where strategic decisions and momentum shifts play a crucial role in determining the outcome. Understanding these dynamics can provide valuable insights for athletes and coaches to enhance performance. In this study, we analyze the role of momentum in tennis matches using data-driven models. To track match progression, we develop a Hierarchical Markov Model, visualizing scoring trends and performance variations in the 2023 Wimbledon Gentlemen’s final. The analysis reveals that Carlos Alcaraz excelled in the 2nd and 3rd sets, while Novak Djokovic performed better in the 1st, 4th, and 5th sets. Additionally, the probability of winning a service point was significantly higher (77.27% for Alcaraz and 79.17% for Djokovic) compared to return points. To investigate the authenticity of momentum, we define it using a weighted approach incorporating technical, psychological, and strategic factors. Bootstrap hypothesis testing (t-statistic = 45.3791, p-value = 0.008) confirms that momentum is not a random phenomenon. Furthermore, logistic regression analysis establishes a strong correlation between momentum and performance. We employ a Long Short-Term Memory (LSTM) model to predict momentum fluctuations, identifying Unforced Errors as the most influential factor. The model effectively forecasts turning points, with an R2 of 0.9501 and RMSE of 0.6701, demonstrating its reliability. Sensitivity analysis and generalizability tests further validate its robustness across different court surfaces and player genders. Our findings offer strategic recommendations for coaches and athletes, emphasizing the importance of minimizing unforced errors, adapting game pace, and capitalizing on momentum shifts. These insights can be instrumental in optimizing match strategies and improving competitive performance.

Keywords

Hierarchical Markov Model, Momentum, Logistic Regression, LSTM, Tennis Strategy