Odds of undefeated season, per Massey.

Submitted by Yeoman on September 26th, 2012 at 9:03 PM

Had some down time tonight so I'm running some numbers off the Massey rating website. Using the odds on each game from their prediction pages (click on each team name on the link to get there), here are the odds on an undefeated regular season in 2012 (not including conference championship games).

I'm guessing the odds are grossly overstated at the high end, because the assumption here is that the current Massey ratings accurately assess each team's strength and the only uncertainty is in the individual game results given those strengths. It doesn't include any uncertainty in the ratings themselves and if I was a bookie there's no way I'd quote odds anywhere near this long on anyone.

  1. Alabama 3:5
  2. OHIO 5:3
  3. Florida State 9:2
  4. Oregon 5:1
  5. Georgia 7:1
  6. Rutgers 8:1
  7. Notre Dame 13:1
  8. Stanford 18:1
  9. Louisiana Tech 19:1
  10. LSU 20:1
  11. Florida 25:1
  12. Louisville 30:1
  13. Texas 31:1
  14. South Carolina 35:1
  15. Cincinnati 35:1
  16. Kansas State 40:1
  17. Ohio State 70:1
  18. Northwestern 130:1
  19. Baylor 185:1
  20. TCU 450:1
  21. Texas Tech 1000:1
  22. Oregon State 1000:1
  23. West Virginia 2000:1
  24. Minnesota 13,000:1
  25. Iowa State 15,000:1
  26. Mississippi State 20,000:1 (that's what happens when Alabama and LSU are on your schedule)

West Virginia is rated ahead of all three Big East teams. Moving from the Big East to the Big 12 probably changed their odds on an undefeated run and a likely trip to the championship game from about 5:1 to...probably not really 2000:1, but still.....

For comparison, Mississippi State is also rated ahead of all three Big East teams. They're also rated just slightly above Ohio State.

For all the griping about the SEC, the real problem with the current system is the enormous advantage it gives to members of the worst BCS conferences, the Big East in particular. Massey's prediction for WVU is 2-7 in conference; if they were still in the Big East they'd be favored in every conference game.

Comments

Yeoman

September 26th, 2012 at 9:20 PM ^

I think you're confusing OHIO with Ohio. It's OHIO UNIVERSITY that's likely to go undefeated, not that other school.

But, to address the point anyway:

Massey ranks West Virginia well ahead of OSU.

Ohio State's favored in all but two of their remaining games (@MSU, @Wisconsin); they don't play anyone currently in the Massey top 20 all year.

West Virginia's only favored in two of their remaining games (Kansas, @Iowa State) and their next seven games are all against teams currently in the Massey top 20.

I can't think of any more telling indication of just how horrible the B1G has been so far this year.

MGoCombs

September 26th, 2012 at 9:51 PM ^

Crazy to me that Alabama is actually favored to go undefeated (if I am reading that correctly). Sure, when you look at any single game, they're favored, but you have to compound those odds over the next 8 games. I shouldn't say I'm surprised but that's a different level of confidence.

Yeoman

September 26th, 2012 at 10:02 PM ^

Here are the odds on Alabama's remaining games:

  1. Mississippi 99%
  2. @ Missouri 96%
  3. @ Tennessee 98%
  4. Mississippi St. 99%
  5. @ LSU 73%
  6. Texas A&M 93%
  7. Western Carolina 100%
  8. Auburn 99%

http://masseyratings.com/team.php?t=74&s=181623

I was thinking they might be getting some extra boost from the Arkansas game--the algorithm doesn't know anything about Arkansas's problems at qb or at hc. So I did a spotcheck and, for what it's worth, the Vegas money line on this week's game is consistent with Massey (+6125 / -11325). I'm guessing they'll shorten the odds on the rivalry game with Auburn, but otherwise this seems about right. They'll be a 7 point favorite at LSU and 20 or more in every other game.

 

LSAClassOf2000

September 26th, 2012 at 10:39 PM ^

If I put $1 on Iowa State, I can pay off most of the remaining balance on  my student loans as well as my wife's. TCU, Oklahoma State, Texas and West Virginia are totally winnable for the Cyclones, right? Wait, what?

turd ferguson

September 26th, 2012 at 10:53 PM ^

I think you're right about there being issues with the way these numbers are calculated. 

Massey probably just found the probability that Team X would win each game and then multiplied them all together.  The problem is that probabilities of winning late season games depend on the outcomes of early season games.  For example, if WVU wins a few in a row, suddenly they'll be seen (by Massey and others) as much more likely to win their remaining games.  If that's what Massey did, then this is kind of a silly exercise, since most/all of these odds should be too long.

I'd need to think more about this, but you'd need a model that learns something about how good a team is if it wins (loses) its next game.  Then you'd have to the probability that it wins that original game and incorporate some conditional probabilities going forward.  Doing that well would be pretty damn complicated.

Yeoman

September 26th, 2012 at 11:07 PM ^

To be clear, Massey only posts individual probabilities for the games; I'm the one that simply mulitiplied them together. Massey has a reputation and they'd never misuse their data the way I just did.

I didn't think anyone would take these odds all that seriously; they get more and more absurd as they get longer (I'll bet you could get 1000;1 on Miss St., though.). If you were a bookie or an insurance underwriter and wanted to do this right you'd need to start with some sort of probability distribution for the strengths instead of a single certain value. You might also need to include some probability of a significant change in strength, like what happened to Arkansas and Utah, losing their quarterback right out of the gate. When you're looking at something like even odds, with Alabama and especially Ohio who probably doesn't have much depth, loss of key personnel is a significant risk.

Or at the very least you could do a sensitivity analysis--what happens to the probability if, say, we increase or decrease the subject school's rating by five points? Massey would probably have some data on how much their ratings change over the course of a season.

I still thought it was interesting, and the relative positions of the teams probably has some merit. it points out in a pretty vivid way how important scheduling is. When you compound over a whole season there's an enormous difference between slogging through the SEC West and playing a Big East or even a B1G Leaders schedule.

Buck Killer

September 26th, 2012 at 11:58 PM ^

Why are these two teams covered and usually over hyped here? Just curious is all. This is the first time I have seen any admittance that WVU never played anyone, and that was forced.

Yeoman

September 27th, 2012 at 9:56 AM ^

so I'll say it a different way: in the Big East West Virginia would be a top 5 team and national title contender; in the Big 12 they'll be an also-ran struggling to get a bowl bid. The move is going to cost them about 4 losses per season, on average.

Any of the schools of similar strength now in power conferences could make that jump in the alternate direction, if they;d move to the Big East. Texas Tech, TCU, Oregon State, Mississippi State--all instant national title contenders if they were in a different conference.

And maybe the biggest problem is that until the postseason we don't, and can't, do anything to distinguish between a mediocre team running the Big East schedule ('09 Cincinnati) and a legitimate national power (I'm not sure this has ever happened, but '05 WVU might have been close--'08 Utah is the best example of this, in a different conference context).

If we're going to do it this way, we need more postseason.

Darth Wolverine

September 27th, 2012 at 10:09 AM ^

I find it funny when buckeye honks claim they will go undefeated and win the "AP national championship," which, let's be honest, doesn't exist anymore. Also, why would the AP award a "national championship" to a team that only played 12 games with no post season whatsoever? That just would not make sense.

rb4kb8

September 27th, 2012 at 10:13 AM ^

Either Rutgers/Louisville.. one of them has a great shot.

My money would be on Louisville... but their schedules are terrible and that final game of the year between Louisville/Rutgers could be a battle of two undefeated teams.

Louisville gets Cincy at home... but has to go to Rutgers.

Rutgers has to go to Cincy, but has Louisville at home.

I'd bet a lot of money either one of them gets it done.