Developing a New Expected Goals Metric to Quantify Performance in a Virtual Reality Soccer Goalkeeping App Called...

Area of Science:

• Sports Science
• Human-Computer Interaction
• Data Science

Background:

• Virtual reality (VR) sports training apps are increasingly popular for athlete development.
• Current VR goalkeeping apps like CleanSheet lack nuanced performance metrics, primarily tracking save rates without considering shot difficulty.
• Aspiring goalkeepers need more sophisticated tools to train effectively using VR technology.

Purpose of the Study:

• To develop a novel shot difficulty metric, termed CSxG (CleanSheet Expected Goals), for VR goalkeeping training.
• To enhance the meaningfulness of performance data derived from VR gameplay.
• To improve the accuracy of goalkeeper performance analysis in VR environments.

Main Methods:

• Developed a regression model combining existing expected goals (xG) models, goalkeeper performance metrics, and psychological research.
• Utilized user save rate data from the CleanSheet VR app as the target variable.
• Incorporated input variables related to ball flight and in-goal positioning, including the required rate of closure (RROC) from Tau theory.

Main Results:

• The developed CSxG model identified the required rate of closure (RROC) as the most significant predictor of goals conceded.
• Validation showed CSxG accurately predicted shot difficulty at the extremes but had lower accuracy for mid-range difficulty scores (0.4-0.8).
• Identified additional factors like build-up play and goalpost size for future model improvements.

Conclusions:

• The CSxG metric offers a more advanced way to assess goalkeeper performance in VR training.
• This research advances predictive modeling in sports analytics, particularly for VR-based training.
• The findings support the potential for enhanced goalkeeper training strategies and performance development through VR technology.