JXC;1614452; said:
Cincy isn't your #1 team. So I wouldn't expect those feelings from you. I have a few friends that love all the math in this stuff. And they said that while UC was 4th in the computers and TCU 5th before today, with Texas and Florida losing, UC would have moved to 2 in the computers, but Texas and Florida would have stayed ahead of TCU and Boise in most of the computers, so TCU's computer average would have been around 4 or 5. It would have been more than enough to move Cincy past TCU in the BCS standings, even with TCU being #2 in the human polls.
Cincy is only in the top ten in the computer rankings if you constrain the rankings to recognize only their absolute performance. That is, to ignore their relative performance against teams by excluding the margin of victory.
On that basis, Ohio State drops from #8 in the Sagarin ratings to #24. Do we really believe that 23 teams are better than Ohio State? That is, do we believe a 5-loss Georgia, Arkansas, or Cal team is better? Four-loss Arizona, Oregon State, USC, or Stanford? Are 31 teams really better than Penn State?
Supposedly, discontinuing margin of victory was supposed to stop top teams from running up the score on weaker teams. That is commendable but, in practice, nothing has changed. Instead, we have a ranking system that fails to include an important piece of information: how much better was one team than another?
There are two immediate problems to this approach. If a good team has a key injury or fails to recover from a tough game the previous week, so that it loses by one point to a team over which it is greatly favored, then...
- The close loss counts just the same as if they had lost by 50 points in a blow-out.
- The losing record of the (lucky) weaker team against even weaker foes is carried throughout the schedule.
This has the effect of magnifying the effects of a silly loss far more than the one game because the win or loss of your opponents versus their opponents affects your rating as an "all or nothing" statistic. It fails to use all of the information provided by a team's performance relative to competitors and results in a biased rank that is not a reliable or valid predictor of team performance.
As an example, Ohio State is ranked so lowly because USC fell apart after Ohio State and we lost to a weak team. The fact that we lost both games by a total of nine points has no relevance to the BCS ranking.
Even after Cincy's win yesterday, Ohio State is ranked higher (#8 vs. #12) on the Predictor ranking scale, which is his most accurate predictor of victories and which takes winning margin into account.