Free access to predictive meta-rankings for college football

News

  • Author: Statistics Views and B. Jay Coleman
  • Date: 17 December 2013

Each week, we select a new article hot off the press and provide free access for a limited period. This week's is from Naval Research Logistics.

To read the article in full, please click on the link below.

Minimum violations and predictive meta-rankings for college football
B. Jay Coleman

Naval Research Logistics, Early View.
DOI: 10.1002/nav.21563

thumbnail image: Free access to predictive meta-rankings for college football

This article presents two meta-ranking models that minimize or nearly minimize violations of past game results while predicting future game winners as well as or better than leading current systems—a combination never before offered for college football. Key to both is the development and integration of a highly predictive ensemble probability model generated from the analysis of 36 existing college football ranking systems. This ensemble model is used to determine a target ranking that is used in two versions of a hierarchical multiobjective mixed binary integer linear program (MOMBILP). When compared to 75 other systems out-of-sample, one MOMBILP was the leading predictive system while getting within 0.64% of the retrodictive optimum; the other MOMBILP minimized violations while achieving a prediction total that was 2.55% lower than the best mark. For bowls, prediction sums were not statistically significantly different from the leading value, while achieving optimum or near-optimum violation counts. This performance points to these models as potential means of reconciling the contrasting perspectives of predictiveness versus the matching of past performance when it comes to ranking fairness in college football.

Related Topics

Related Publications

Related Content

Site Footer

Address:

This website is provided by John Wiley & Sons Limited, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ (Company No: 00641132, VAT No: 376766987)

Published features on StatisticsViews.com are checked for statistical accuracy by a panel from the European Network for Business and Industrial Statistics (ENBIS)   to whom Wiley and StatisticsViews.com express their gratitude. This panel are: Ron Kenett, David Steinberg, Shirley Coleman, Irena Ograjenšek, Fabrizio Ruggeri, Rainer Göb, Philippe Castagliola, Xavier Tort-Martorell, Bart De Ketelaere, Antonio Pievatolo, Martina Vandebroek, Lance Mitchell, Gilbert Saporta, Helmut Waldl and Stelios Psarakis.