The Evaluation of the Team Performance of MLB Applying PageRank Algorithm

Abstract

Background. There is a weakness that the win-loss ranking model in the MLB now is calculated based on the result of a win-loss game, so we assume that a ranking system considering the opponent’s team performance is necessary. Objectives. This study aims to suggest the PageRank algorithm to complement the problem with ranking calculated with winning ratio in calculating team ranking of US MLB. Methods. PageRank figure is calculated by using the result of 4,861 matches in the 2017 season (2,430 matches) and 2018 season (2,431 matches) in the MLB. Results. There is a difference between ranking calculated in PageRank and ranking calculated with winning ratio both in the 2017 season and 2018 season, and there is a difference in performance per each district due to comparing performance per each league and district. In addition, as a result of calculating the predictive validity of PageRank and winning ratio ranking, it turns out that the ranking calculated with the PageRank algorithm has relatively high predictive validity. Conclusion. This study confirmed the possibility of predictive in the US MLB by applying the PageRank algorithm.