Monday, December 08, 2003

BCS Follies

Let me start off by saying that I had Texas Football season tickets for the six years that I lived in Austin. Good ones too, between the 45's on the east side of the stands. This may lead to a certain bias against Oklahoma Football, although a much bigger one against the Aggies. My current perspective from the Great White (for today, anyway) North hasn't turned me into a Pitt fan, or even a Steelers fan (that's a different comment) so Texas is still my default team to follow.

That said, what on earth has the BCS wrought? While there are many bad things about the BCS system[1], this week's revelations will surely show it for the fraud that it is. Well, they won't because there is too much money at stake, but they should. Having the consensus #1 pick in both polls miss the national championship does seem a bit daft. It certainly wouldn't have happened before the institution of the BCS, since Oklahoma either would have tried to play on Saturday night instead of resting on their laurels, secure in the knowledge that they had made the big game already.

What this really drives home, however, is the incredible over reliance of the BCS on strength-of-schedule. It not only appears in directly in the average, but it is also explicitly included in each computer ranking, and implicitly included in the human polls, and even kind of shows up in the quality win column. While the computer poll algorithms are confidential, it wouldn't surprise me to find out that adding up all the strength of schedule contributions to the BCS rankings would show that it accounts for over 50% of the total ratings. Since the computer polls have to agree not to look at margin of victory, or the other things on which human pollsters rely, they are pretty much constrained to W/L record and SOS.

According to the charts over at ESPN, simply allowing strength of schedule to remain implicit (i.e. subtracting out the explicit SOS rankings) leaves OSU and USC playing for the title instead of LSU. It's still not the human poll result, but it seems a little closer to reason. At least the consensus number one would be playing for the championship. If we leave out the "quality win" kludge as well and just look at the poll averages (human and computer) then the result is finally correct.


[1] The worst thing about the BCS is that it pretty much destroys the bowl system it is trying to save. Suddenly only one college bowl game has much national interest at all, and it isn't even played on New Year's Day. I use to religiously nurse my hangover on January 1st in front of 12-14 hours of solid football -- like a squirrel stocking up for the long cold season ahead. For the past few years, I may occasionally glance at a game, but that's it. For me, the BCS has pretty much pushed the bowl games into irrelevance in a way that a playoff system just wouldn't.