Problem C of the 2024 COMAP Mathematical Contest in Modeling (MCM) challenged participants to analyze play-by-play Wimbledon tennis data to uncover insights regarding swings of “momentum” in tennis (i.e., swings of competitive advantage in matches). As part of our solution, we develop (1) a Markov-chain model for the lead that one player possesses over another, (2) a numerical model for the momentum one player has built up in the match, based on the lead function, and (3) a machine learning model to capture the probability of a given play being a momentum shift, given the current lead, momentum, and specific aspects of the play. We then analyze the performance, generalizability, and sensitivity to various inputs of these models and conclude with a letter to tennis coaches based on our results.

Our report was awarded Honorable Mention, placing us in the top ~20% of 10,000+ teams competing in the 100 hour competition.