A Study of Momentum Analysis Based on CRITIC and SA-LGBM

Accurately quantifying and understanding momentum in competition is an important challenge for both athletes and coaches. In this study, we developed three key models by analyzing match data: a momentum assessment model, a momentum-success correlation model, and a momentum trend prediction model. We found that momentum plays an important role in match outcomes, especially in terms of break points, serve advantage, point advantage, and unforced errors. The CRITIC weighting method is applied to assign momentum to each metric, and the results are presented through a variety of visualization methods. Ultimately, we utilize the Cumsum algorithm to detect the close correlation between momentum scores and match outcomes and select the LGBM model for prediction via the SA algorithm, achieving excellent results and providing insights for quantifying and understanding momentum in matches.