抄録
Volleyball features a unique rule known as rotation, and while previous studies have quantified its importance, they
often fail to consider the serving phase, that is, whether a team has the right to serve. In this study, we used machine learning predictions to quantitatively analyze the importance of rotation while accounting for the serving phase. We collected data using Volley Station from 180 matches in the 2023–24 V.LEAGUE DIVISION 1 MEN's League (V1 Men) and 82 matches in the 2022–23 V.LEAGUE DIVISION 2 MEN's League (V2 Men). A machine learning-based win-loss prediction model was employed for the analysis. The explanatory variables were the scoring percentages in 12 distinct situations, each combining different phases and rotations. The outcome variable was the match result, coded as 1 for a win and 0 for a loss. We utilized three types of predictive models, logistic regression, linear discriminant analysis, and Light GBM, to calculate prediction accuracy and assess the importance of serving phase and rotation. Our analysis revealed that in the V1 Men's League, break phase rotations B-S6 and B-S5 were especially important. In these rotations, the opposite hitters were in front-row positions, and differences in blocking ability were pronounced. Conversely, in the V2 Men's League, side-out phase rotations S-S6, S-S5, and S-S2 were particularly significant. In these rotations, the ability to execute reception attacks showed a marked difference between winning and losing teams.
