Tuesday, November 9, 2010

BCS Championship Game Odds Model

Our BCS Model currently predicts the following chances of these individual teams playing in the BCS Title game -



Oregon
76.89%
TCU
75.53%
LSU
12.96%
Nebraska
11.29%
Auburn
10.63%
Boise State
8.86%
Oklahoma State
1.76%
"Other"
2.08%

The odds of specific match-ups is as follows -



TCU-ORE
56.70%
LSU-TCU
6.14%
AU-ORE
5.81%
NEB - TCU
5.17%
LSU-ORE
4.90%
BSU-ORE
4.45%
NEB - ORE
4.43%
BSU-TCU
3.42%
TCU-AU
3.26%
NEB - LSU
0.87%

OK ST
- TCU
0.84%
LSU-AU
0.61%

OK ST
- ORE
0.60%
NEB - AU
0.51%
BSU-AU
0.35%
BSU-NEB
0.31%
BSU-LSU
0.27%

OK ST
- LSU
0.18%

OK ST
- AU
0.08%

BSU-OK ST
0.06%
"Other"
1.04%


A couple of Model notes/assumptions –

- Auburn is still thought to have a very poor chance at Alabama per the prediction experts (Sagarin has Alabama 5th, Auburn 11th in his predictor).

- I believe LSU is better positioned to go to the title game if they win out than Boise State. LSU is 4th in the computer polls to Boise’s 5th, and LSU has a cumulative SOS per the NCAA of 5th nationally, while Boise is ranked 58th.

- I also suspect, and this could be controversial, that either Nebraska or Oklahoma State, should they win out, will jump an unbeaten Boise State team. Oklahoma State is currently un-loved by poll voters, but ranked 6th in the computers just barely behind Boise State (.770 to .790). If the Cowboys win out they will have to beat Oklahoma, and have a “revenge” win against Nebraska. I think, under these circumstances, the computers will love them, and the human voters will come around. Right now Oklahoma State’s cumulative NCAA SOS is 27th nationally, and that is WITHOUT accounting for a Big 12 Championship game, which will put them in the top 10 SOS.

- I don’t see either Ohio State or Wisconsin having a chance. OSU is 15th in the computers, and their cumulative SOS is 46th. Wisconsin is 10th in the computers, with a cumulative SOS of 58th.

- In that 1.04% of “other” lies Stanford, who would depend entirely on Oregon and a couple of others losing.

So there you have it. The model says a BCS title game between Oregon and TCU is more likely than not. I also think this could be a compelling match-up of the offense of Oregon against the defense of TCU.

UPDATE: In response to reader comments, I would add that the reason TCU-Oregon is such an overwhelming favorite is based mostly upon the calculations that Auburn has such a poor chance of running the table. Sagarin has them as a 9 point underdog to 'Bama, while statfox puts as a 13 point dog. If they lose to 'Bama as expected, then play a weak SEC East team (are there any others?) I don't think theymake it.

It appears LSU, on the other hand, could make it even without playing for the SEC title. However the odds are pretty slim. The most likely LSU matchup, with TCU, would only occur if Oregon and Auburn lose, and only comes in at 6%. But look at it this way - LSU will end up with one of the nation's toughests SOS's (15th right now in Sagarin, the toughest of his top 10). If Oregon and Auburn somehow manage to lose, who else is going to benefit? The computers will love them, and human pollsters seeking to avoid a Boise-TCU title game will elevate them. In the unlikely event both Oregon and Auburn lose, I can't think it will be anyone else BUT LSU.

Also let's not get get too caught up in the 2nd six most likely match-up's - they are all about 5% chances, so getting wrapped up in their rankings is pointless.

3 comments:

Anonymous said...

The "Fly" in LSU's ointment?

For LSU to win the West, Auburn would have to lose to both Alabama and Georgia.

Would an AU loss to a sub-.500 UGA team weaken LSU's SOS (and computer ranking) enough to offset the positive from Alabama beating Auburn?

Amos said...

I think you hit the nail on the head with that. I think LSU only gets to the BCS CG if they win the SEC, and I think the chances of Auburn losing both those games isn't high. So LSU might be a great 1-loss team, but they won't get voted into the top 2 by most pollsters come final pre-bowls poll.

Anonymous said...

I think its fascinating that the second most likely matchup is between the current #3 and #5...
and a 3 v. 8 game is almost as likely to happen as a 1-2 tilt.
Just goes to show how crazy this season has been--it's even screwing with "the numbers."