Tennis Winner Prediction based on Time-Series History with Neural Modeling

Tennis is one of the most popular sports in the world. Many researchers have studied in tennis model to find out whose player will be the winner of the match by using the statistical data. This paper proposes a powerful technique to predict the winner of the tennis match. The proposed method provides more accurate prediction results by using the statisti- cal data and environmental data based on Multi-Layer Perceptron (MLP) with back-propagation learning al- gorithm. Due to the growth of sport betting, predictions are widely used in many kinds of sports, especially tennis. The ten- nis prediction model is created to evaluate the chance of winning and the expected length of the match that players will face. Most people believe that the first serve person in the set has more advantage than another be- cause most of the games often go like that so the first serve affect to the games' score (2). Additionally, lots of players always make fault in the first serve and do bet- ter in the second serve so second serve might affect to the games' score too. Nevertheless, the first serve and the second serve affect to the games' score but there is another thing, that might be refuting an advantage of serves, it is strongly returns of serve. Moreover, the sur- face characteristics also affect to the players, e.g., some players perform better on grass but they may get worse on clay.