题目：How competitive Is competition？A Bayesian View of Parity Over Time

汇报人：胡冠宇

会议时间：2026年5月28日(周四)      13:30—15:30

地点： 综合楼601会议室

报告人简介：现任密歇根州立大学副教授，研究方向主要包括贝叶斯非参数方法和体育数据分析。曾担任 ASA Statistics in Sports Section 主席、ISBA East Asia Chapter 项目主席，并担任 The Annals of Applied Statistics、Biometrics 和 Journal of Quantitative Analysis in Sports 的副主编。他也是 American Soccer Insights Summit 组织委员会成员。他的研究致力于发展灵活的贝叶斯方法，用于建模体育表现、竞争结构与决策过程。

摘要：How competitive is a competition, really? In professional sports leagues, parity mechanisms aim to keep teams evenly matched, yet observed rankings often reflect sampling variability as much as true differences in ability. This raises important statistical questions: how can competitive balance be quantified, how does it evolve over time, and what rank tier structure can be learned from the data?In this talk, I present a Bayesian framework that treats competitive balance as a model based quantity. The model represents latent competitor abilities within a time varying geometric constraint whose scale provides an interpretable measure of parity. This structure also induces a data driven rank tier system, allowing competitors to be grouped when differences are not credibly distinguishable. To capture temporal dynamics, I introduce a hidden Markov mixture structure that allows competitive regimes and tier configurations to persist and evolve over time. Applications to professional sports leagues illustrate how the framework can be used to study parity, ranking uncertainty, and the persistence of competitive tiers.