We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

physics

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Physics > Physics and Society

Title: A network-based ranking system for American college football

Abstract: American college football faces a conflict created by the desire to stage national championship games between the best teams of a season when there is no conventional playoff system to decide which those teams are. Instead, ranking of teams is based on their record of wins and losses during the season, but each team plays only a small fraction of eligible opponents, making the system underdetermined or contradictory or both. It is an interesting challenge to create a ranking system that at once is mathematically well-founded, gives results in general accord with received wisdom concerning the relative strengths of the teams, and is based upon intuitive principles, allowing it to be accepted readily by fans and experts alike. Here we introduce a one-parameter ranking method that satisfies all of these requirements and is based on a network representation of college football schedules.
Comments: 15 pages, 3 figures
Subjects: Physics and Society (physics.soc-ph); Disordered Systems and Neural Networks (cond-mat.dis-nn); Data Analysis, Statistics and Probability (physics.data-an)
Journal reference: Journal of Statistical Mechanics, P10014 (2005)
DOI: 10.1088/1742-5468/2005/10/P10014
Cite as: arXiv:physics/0505169 [physics.soc-ph]
  (or arXiv:physics/0505169v4 [physics.soc-ph] for this version)

Submission history

From: Juyong Park [view email]
[v1] Tue, 24 May 2005 17:07:26 GMT (56kb)
[v2] Wed, 25 May 2005 15:55:38 GMT (56kb)
[v3] Wed, 3 Aug 2005 18:34:21 GMT (53kb)
[v4] Mon, 31 Oct 2005 19:11:51 GMT (59kb)

Link back to: arXiv, form interface, contact.