Beyond Skill: Predictive Modeling with Individual and Team Attributes in League of Legends

AbstractThe goal of this study is to explore the predictive capability of several psychosocial variables, such as personality and group cohesion, towards determining multiplayer online battle arena game outcomes - namely diversity, cohesion, and resilience, on collective performance. Our study finds that measures of individual and team perceptions of qualities provided a useful precursor for match victory. Using individual-level attributes, our cohesion survey questions provided the highest predictive value, and higher levels of perceived cohesion were associated with higher victory odds. In light of our results, we discuss the implications of using behavioral data derived from online games and opportunities for future large-scale game data collection.


Return to previous page