This work describes a collaboration with a single collegiate volleyball team to leverage existing data to examine the potential of the plus/minus metric for player evaluation. Historically, volleyball players have been evaluated through a series of single skill metrics (e.g., number of aces per set and hitting percentage). The advantages of the plus/minus lie in the limited amount of information needed for its calculation (e.g., court presence and scoring) combined with its ability to fuse together both measured and unmeasured contributions. Unfortunately, the primary collection tool (Statcrew) for National Collegiate Athletic Association (NCAA) Women`s Volleyball, does not record the movement of the Libero, resulting in incomplete court presence information for a large percentage of plays. This paper introduces methodology to recover court presence information from standard play-by-play data. The recovery is in the form of a posterior distribution of player presence, which can then be used to not only calculate the plus/minus metric but also quantify the uncertainty of the metric due to the incomplete information. Although the presented methods and results were derived from a collaboration with a single team, the data source and methodology can be extended to multiple teams.
© Copyright 2018 Journal of Sports Analytics. IOS Press. Alle Rechte vorbehalten.