among those who run the 6 computers that account for 1/3 of a team’s BCS rating. Billingsley does not have an especially strong mathematical or scientific background. What really sets his system apart, though, is an algorithm that is unique among the BCS computers. Various observers have noted that it seems to work more like a poll voter than a computer program.
This earlier post notes that Billingsley‘s rankings are tossed out more than any other system’s as the highest or lowest for a particular team. As can be seen on this site’s computer page, the key difference of Billingsley‘s algorithm is that the strength of schedule of a team does not change as that team’s earlier opponents win or lose games after the matchup. This can be seen on the ratings page of Billingsley‘s site by noting that, with very few exceptions, a team’s BCS rating does not go up or down during a bye week. Indeed, Billingsley does not directly consider opponents’ win-loss records – rather, their ranks and ratings are taken into account.
Head-to-head matchups are very important in Billingley‘s system. As an example, early in the 2009 season, Boise State‘s BCS rating did increase during a bye week. Under Billingsley‘s algorithm, as an undefeated team, Boise State must have been kept ahead of Oregon, a team it defeated previously. Another unique aspect is that, as noted in this article, the system is the only one that takes into account preseason polls.
Billingsley publishes conference and strength of schedule rankings on a separate page of his site. His first BCS rankings of 2009 were published on September 8, and as an observant reader will notice, the final rankings of the previous season were considered. Billingsley writes about his ranking system on this page.