Predicting Table Tennis Tournaments: A comparison of statistical modelling techniques

International Journal of Racket Sports Science

Hendrik van der Wurp¹  Andreas Groll¹  Jan Lennartz¹ 

[1] TU Dortmund University, Department of Statistics, Dortmund;

关键词: Tournament Analysis;    Random Forest;    Statistical Learning;    Table Tennis;    LASSO Regression;   

DOI  :  10.30827/Digibug.73877

来源: DOAJ

【 摘 要 】

Every year, at least one of four important recurring table tennis tournaments takes place, where top players compete. Those tournaments are the World Table Tennis Championships, the Table Tennis World Cup, the Olympic Games and the ITTF World Tour. In other areas of sports, it is common to analyse major tournaments and predict future ones (see, e.g., Groll et al., 2018, for football). This work aims to bring this aspect of analysis to the world of table tennis by conducting recent holdings of the Men’s World Cup and the Grand Finals of the Men’s ITTF World Tour. There are two main goals: 1) to compare different modelling techniques on historic tournaments to find the model with the best predictive performance, and 2) to understand which factors are important for good predictions. The results show that it is indeed possible to apply statistical machine learning methods on table tennis tournaments for prediction with a correct classification rate of around 75% by a random forest and 74% by a penalized generalized linear logit model. Even though both models based their predictive power mainly on the official table tennis rankings and points, variables like age, playing hand or individual strength were important factors as well.

【 授权许可】

Unknown