Tennis Match Result Forecast And Player Analysis

With the continuous breakthrough of Chinese athletes in the international tennis court,tennis is favored by more and more fans in Our country,and the development of the tennis betting market is increasingly prosperous.Both fans and betting companies are very interested in predicting the results of tennis matches.This paper provides theoretical basis and reference value for fans and betting companies by analyzing the influencing factors of tennis matches and predicting the results of tennis matches.Firstly,this thesis studies the influencing factors of tennis match.Considering the three links of serving,receiving and scoring,sixteen technical indexes are selected to construct the technical index system.Three common factors affecting player performance are extracted by factor analysis method,which are service performance factor,service return performance factor and service stability factor.The top 16 players in the world are rated on three common factors,divided into three grades and evaluated according to their strength.Secondly,markov model based on ranking system is used to predict tennis match results.The probability of a player winning a set,tiebreak,set and match is recursively calculated based on the probability of a player scoring a serve.The ranking system is improved,and the probability of serving points is estimated by using the improved ranking system,and the influence of different types of courts and the internal differences of serving and receiving links are incorporated into the model,and the ranking system and model are revised.The average prediction percentage of the improved Markov model is 0.68822,which is 0.65% higher than that of the original Markov model,and the deviation is also reduced.Then,the support vector machine model,logistic regression model and multi-layer perceptron model in machine learning are used to predict the outcome of tennis matches.The best model is the support vector machine model,which has a prediction accuracy of73% in the test set.The ablation study of each feature shows that the most important feature in determining the outcome of a match is the player’s historical record.Finally,tennis betting.Using XGBoost models and odds data to place bets on selected races,the average return on investment was 20% by placing bets on 35% of the races.The innovation of this paper lies in the combination of Markov model and ranking system,forming a new tennis modeling method.The research results of this paper can not only provide suggestions for players,but also have certain reference significance for football fans and betting companies.