A1608
Title: Beyond univariate analysis: Detecting mean and variance shifts across North American sports leagues and baseball history
Authors: Mena Whalen - Loyola University Chicago (United States) [presenting]
Brian Mills - University of Texas at Austin (United States)
Gregory Matthews - Loyola University Chicago (United States)
Abstract: Recent methodological developments in multivariate change point analysis are applied to examine structural shifts across two distinct domains of sports economics. The method identifies changes in both level (mean) and variance simultaneously, offering a more complete picture of structural change than prior univariate approaches. In the first application, competitive balance in North American professional sports leagues is tracked by simultaneously examining nine competitive balance measures spanning game uncertainty, playoff uncertainty, and consecutive season uncertainty. This approach identifies the specific measure(s) driving each estimated change point and captures changes to the variance structure of competitive balance. Results suggest that rival leagues, labor supply increases, expansion, and changes to playoff structures are the factors most commonly concurrent with structural breaks. In the second application, the same framework identifies distinct historical eras in Major League Baseball using team-level performance statistics from the late 1800s through 2021. The method recovers well-established eras including the Dead Ball Era, Integration Era, Steroid Era, and Post-Steroid Era, while also pinpointing periods of team-level dynastic dominance and collapse.
