Robert Seidl, Patrick Lucey
A recent trend in machine learning is to utilize interpretable techniques such as counter-factual analysis to explain predictions of individual events. Such techniques are powerful in sports where it can be used to answer the impact of a play or event on the overall outcome of the match by framing it as a “what-if” questions (i.e., if a player win/loses the next point – how does that change the likelihood of her winning the game/set/match?). In this paper, we present a counter-factual method for women’s tennis that first automatically highlights the key moments in a match using our “leverage”, “clutch” and “momentum” metrics. Not only can our approach highlight important moments before they occur in an automatic fashion, it can also link player behaviors at a season level which shines a light on their tendencies in key moments.