Set-level strategic and psychological momentum in best-of-five matches in professional tennis

We provide a theoretical and empirical analysis of set-level strategic and psychological momentums in a sample of 8,193 ATP best-of-five-set tennis matches where players have asymmetric abilities. Theoretical predictions match observed Set 1 outcomes, the only set with no set-level psychological momentum, but do not match any other sets. We find evidence of psychological reversal: in every set following a non-tie net score, the winner of the previous set underperforms compared to theoretical predictions.