Abstract
Background
This study developed a sequential winning-percentage prediction model for badminton competitions using the expert system sequential probability ratio test (EXSPRT), aiming to calculate the difficulty of each event within a match and establish the initial prior probability.
Methods
We utilized data from 100 men's singles matches (222 games) held by the Badminton World Federation (BWF) in 2018 to evaluate event difficulty across six models for each determining factor. For setting the initial prior probability calculation method, 30 men's singles matches (74 games) organized by the BWF in 2019 were randomly selected. The odds for these matches were obtained from www.oddsportal.com.
Results
The efficacy of the six models was assessed based on application rates (15%, 20%, 25%, and 30%) of the collected odds, with the initial prior probability reflecting 25% of the odds chosen owing to its superior validity.
Conclusions
This research yielded six sequential winning percentage prediction models capable of offering real-time predictions during matches in badminton competitions by leveraging EXSPRT. These models enhance spectator engagement and provide foundational data for developing similar prediction models for other sports. Future research should focus on developing a program to identify the most effective model among the six and implement it practically.