Publication Date: September 2002
Revision Date: March 2005
The results in this paper show that various college football ranking systems have useful independent information for predicting the outcomes of games. Optimal weights for the systems are estimated, and the use of these weights produces a predictive system that is more accurate than any of the individual systems. The results also provide a fairly precise estimate of the size of the home ﬁeld advantage. These results may be of interest to the Bowl Championship Series in choosing which teams to play in the national championship game. The results also show, however, that none of the systems, including the optimal combination, contains any useful information that is not in the ﬁnal Las Vegas point spread. It is argued in the paper that this is a fairly strong test of the eﬀiciency of the college football betting market.
Football rankings, Predictive information
JEL Classification Codes: C52
Published in Journal of Sports Economics (February 2007), 3-18