Tennis Data

Forecasting professional tennis players winning matches has a wide range of practical applications. We introduced a new approach to measure and combine strategic and psychological momentum using the entropy weight method and the analytic hierarchy process, and test its effectiveness. Using data from Wimbledon Championship 2023, we then constructed a support vector machine (SVM) model to predict the turning point and winner of each point, and we optimized it using particle swarm optimization (PSO). Our model achieved a significant level of accuracy (96.09\% turning point and 83.52\% predicting winner) and performs well in different courts and players. Furthermore, we compare its performance with commonly utilized predictive models, including ARIMA, LSTM and BP network, and find that our model exhibits higher accuracy than other existing models on predicting the point winner. Our research can be used to calculate odds in tennis matches and provide advice to coaches.预测职业网球选手的赛事获胜结果，具备广泛的实际应用场景。本研究提出一种全新方法，借助熵权法（Entropy Weight Method）与层次分析法（Analytic Hierarchy Process）对战略动量与心理动量进行量化与融合，并验证了该方法的有效性。依托2023年温布尔登网球锦标赛（Wimbledon Championship 2023）的赛事数据，本研究构建了支持向量机（Support Vector Machine, SVM）模型以预测每一分的转折点与获胜方，并采用粒子群优化（Particle Swarm Optimization, PSO）算法对模型进行优化。所提模型取得了出色的预测性能：转折点预测准确率达96.09%，单分获胜方预测准确率达83.52%，且在不同赛场与不同选手的赛事场景下均表现稳定。此外，本研究将所提模型与自回归积分滑动平均模型（AutoRegressive Integrated Moving Average, ARIMA）、长短期记忆网络（Long Short-Term Memory, LSTM）以及反向传播神经网络（Back Propagation Network, BP网络）等主流预测模型开展性能对比，结果显示所提模型在单分获胜方预测任务上的准确率优于所有对比模型。本研究可应用于网球赛事的赔率计算，并可为教练员提供战术指导参考。