Propensity Score Matching for Evaluating Flashbang Effects in Counter-Strike

The context-driven nature of spatial-temporal sports data can make analysis of player actions difficult. We explore the Esports Trajectories and Actions (ESTA) dataset to evaluate flashbang grenade usage of professional Counter-Strike: Global Offensive (CSGO) players. We use propensity score matching, a non-parametric preprocessing technique that aims to enable causal inference in obser-vational studies, to assess flashbang grenade effectiveness. This approach highlights the nuances of the dataset and has potential for application to other spatial-temporal datasets.