Nonparametric Estimation in the Dynamic Bradley-Terry Model

We propose a time-varying generalization of the Bradley-Terry model that allows for nonparametric modeling of dynamic global rankings of distinct teams. We develop a novel estimator that relies on kernel smoothing to pre-process the pairwise comparisons over time and is applicable in sparse settings where the Bradley-Terry may not be fit. We obtain necessary and sufficient conditions for the existence and uniqueness of our estimator. We also derive time-varying oracle bounds for both the estimation error and the excess risk in the model-agnostic setting where the Bradley-Terry model is not necessarily the true data generating process. We thoroughly test the practical effectiveness of our model using both simulated and real world data and suggest an efficient data-driven approach for bandwidth tuning.

[1]  Filip Radlinski,et al.  Active exploration for learning rankings from clickthrough data , 2007, KDD '07.

[2]  Devavrat Shah,et al.  Rank Centrality: Ranking from Pairwise Comparisons , 2012, Oper. Res..

[3]  L. R. Ford Solution of a Ranking Problem from Binary Comparisons , 1957 .

[4]  S. Stigler Citation Patterns in the Journals of Statistics and Probability , 1994 .

[5]  Alan Agresti,et al.  Categorical Data Analysis , 2003 .

[6]  Mark E. GLICKMAN,et al.  A State-Space Model for National Football League Scores , 1998 .

[7]  R. A. Bradley,et al.  RANK ANALYSIS OF INCOMPLETE BLOCK DESIGNS , 1952 .

[8]  E. Zermelo Die Berechnung der Turnier-Ergebnisse als ein Maximumproblem der Wahrscheinlichkeitsrechnung , 1929 .

[9]  L. Fahrmeir,et al.  Dynamic Stochastic Models for Time-Dependent Ordered Paired Comparison Systems , 1994 .

[10]  Larry A. Wasserman,et al.  Time varying undirected graphs , 2008, Machine Learning.

[11]  David Firth,et al.  Statistical modelling of citation exchange between statistics journals , 2013, Journal of the Royal Statistical Society. Series A,.

[12]  Benjamin S. Baumer,et al.  How often does the best team win? A unified approach to understanding randomness in North American sport , 2017, The Annals of Applied Statistics.

[13]  R. A. Bradley,et al.  RANK ANALYSIS OF INCOMPLETE BLOCK DESIGNS THE METHOD OF PAIRED COMPARISONS , 1952 .

[14]  Mark E. Glickman Paired Comparison Models with Time-Varying Parameters , 1993 .

[15]  C. Varin,et al.  Dynamic Bradley–Terry modelling of sports tournaments , 2013 .

[16]  RaghavanPrabhakar Probabilistic construction of deterministic algorithms: approximating packing integer programs , 1988 .

[17]  Ulrike von Luxburg,et al.  A tutorial on spectral clustering , 2007, Stat. Comput..

[18]  Cristiano Varin,et al.  The ranking lasso and its application to sport tournaments , 2012, 1301.2954.

[19]  Samuel L. Ventura,et al.  nflWAR: a reproducible method for offensive player evaluation in football , 2018, Journal of Quantitative Analysis in Sports.

[20]  Matthias Grossglauser,et al.  Pairwise Comparisons with Flexible Time-Dynamics , 2019, KDD.

[21]  Yi-Ching Yao,et al.  Asymptotics when the number of parameters tends to infinity in the Bradley-Terry model for paired comparisons , 1999 .