I see it as my duty—since no one in the media seems to want to do it—to correct misconceptions about how the BCS computer algorithms work. While it may be difficult to get into details, a basic concept that all the BCS computer formulas must follow is this: margin of victory is not allowed to be used as a factor, period.
So when Andrew Bagnato writes in College Football Today,
With the first Bowl Championship Series standings due out Sunday, was No. 6 Southern California trying to send a message to the voters and the computers in a 69-0 rout of Washington State? Maybe.
He's inadvertantly spreading disinformation. The computers won't be impressed by the rout, no more than they'd be impressed by a 3-0 win by the Trojans. They won the game, and that's all they see.
In fact, some of the formulas might penalize USC for winning the game. If they use overall (average) strength of schedule in their algorithm rather than considering each game individually (like my Success power rating, which doesn't use score margin, and where teams aggregate points for each win), USC's schedule strength may decline and make them rate slightly lower.
Some of the formulas have had this effect in the past, though I'm not certain they still do—I don't want to spread disinformation of my own in this campaign season, but it's a common feature of many power ratings.
The really sad thing is, it probably will send a message to the voters. Not that a huge shutout win shouldn't send a message; I believe in using score margin to an extent when evaluating team performance. But let's put this one in context: they beat the 119th best team in Division I-A 69-0. Just because Washington State is a BCS conference school doesn't make the win impressive. But when voters see USC crush a team 69-0, the Trojans get a lot more credit than other schools would, mostly because no one pays attention to scores like these:
- California 66, Washington State 3
- Oregon 63, Washington State 14
- Oregon State 66, Washington State 13
From these results, it's clear no one should be impressed that a top ten team could score 69 on the Cougars, nor that they shut them out. It's almost the expected result. Using my Strength power rating, which *does* use score margin for it's results, the predicted result was USC 55, Washington State 0.
Therefore, a score-margin-based power rating might be somewhat impressed by USC's win. The voters, however, should note that everyone in the Pac-10 was scoring 60+ on the Cougars, so high-ranked USC should do the same. And certainly, none of the BCS computers will care at all.
It will be interesting to see if USC jumps anyone else due to beating one of the worst teams in Division I-A by a score that should come as no surprise to anyone. In the AP poll USC was 6th and Florida—idle this week—5th. The Harris poll finds USC tied with Oklahoma, which is a joke considering Oklahoma's résumé, which was strengthened further with a win over Kansas. The Sooners should pull ahead this week, unless USC's win mistakenly sways the Harris voters more.
Comments