Ranking in the generalized Bradley-Terry models when the strong connection condition fails

TL;DR

This paper introduces an improved penalized likelihood approach for ranking objects in generalized Bradley-Terry models when the strong connection condition fails, ensuring robust rankings even in challenging comparison scenarios.

Contribution

It extends the $ ext{ε}$-singular perturbation method to generalized models and provides conditions for the existence and uniqueness of penalized MLEs.

Findings

Ranking remains robust across different ε values.

The method successfully applies to NFL data.

Conditions for MLE existence are established.

Abstract

For nonbalanced paired comparisons, a wide variety of ranking methods have been proposed. One of the best popular methods is the Bradley-Terry model in which the ranking of a set of objects is decided by the maximum likelihood estimates (MLEs) of merits parameters. However, the existence of MLE for the Bradley-Terry model and its generalized models to allow for tied observation or home-field advantage or both to occur, crucially depends on the strong connection condition on the directed graph constructed by a win-loss matrix. When this condition fails, the MLE does not exist and hence there is no solution of ranking. In this paper, we propose an improved version of the $\varepsilon$ singular perturbation proposed by Conner and Grant (2000), to address this problem and extend it to the generalized Bradley-Terry models. Some necessary and sufficient conditions for the existence and uniqueness of the penalized MLEs for these generalized Bradley-Terry-$\varepsilon$ models are derived. Numerical studies show that the ranking is robust to the different $\varepsilon$. We apply the proposed methods to the data of the 2008 NFL regular season.