Machine learning-based optimization of contract renewal predictions in Korea Baseball organization.

The Korea Baseball Organization (KBO) introduced the foreign player system in 1998 to enhance league competitiveness. In 2014, the number of foreign players increased to three per team. Since then, the ten KBO teams have routinely included two foreign pitchers and one batter. While the performance of foreign players significantly impacts the post-season qualification of the team, the contract renewal rates for pitchers and batters are only 34 % and 36 %, respectively. Therefore, a method that can aid in the contract renewal decision can help teams recruit high-caliber foreign players, improve their performance, raise the level of KBO league, and provide an enjoyable experience for baseball fans. In this study, we use machine learning methods to predict the contract renewal decision and compare the prediction performances of various models. We use data on foreign player performances in the Minor League Baseball and Major League Baseball immediately prior to joining KBO, in KBO upon joining, and player image data. By comparing the accuracy, area under the receiver-operating characteristic curve, and precision of prediction results based on performance in each league, we find that performance in KBO plays a significant role in improving the prediction. Additionally, a post-hoc analysis of batters reveals a gradual decline in the performance level of foreign batters who succeeded in KBO, which is found to be related to the results of international baseball tournaments. In conclusion, the proposed approach to player performance evaluation and contract renewal decisions can contribute to the long-term success of the teams and league.