Should Vegas Be In Charge Of This Mess?

by PB @ BON on Oct 12, 2006 12:25 PM CDT in Football Comment 12 comments

So, remember yesterday, when I appeased a certain Washington blogger who just wanted a little shout out from a Horns blog?

Turns out I was the big winner of that exchange, as I wasn't aware of Dan Steinberg's blog prior to yesterday. And it's terrific.

Steinberg's been burning some serious brain power trying to figure out the best way to rank college football teams. Much like what AR's got going in this weeks' Two Cents series.

Steinberg's idea today? Turn it over to Las Vegas.

Turns out I'm more than a year late to this party. Las Vegas Sports Consultants, the leading consultant for Nevada sports books, started publishing their own OddsMakers Top 25 last year...

I talked to LVSC senior oddsmaker Mike Seba a few minutes ago, and told him that I thought the oddsmakers poll would likely be more accurate than the AP or the coaches or the Harris people.

"Good man," he said. "You're right. When you look at some of the polls from year to year, it just makes you sick sometimes to see where some of these teams are rated. Whereas we know if you matched two teams up on a neutral field, who would be the favorite. It depends on who they played, and where they played, and when they played. Our livelihood depends on it, whereas these guys voting in the AP, they may care less, they could spend five minutes looking at it.

Ho ho! Now this is interesting. And about the sexiest idea I've heard in well over a year.

Steinberg continues:

I asked another LSVC oddsmaker, Sean Van Patten, whether the OddsMakers Top 25 would be a better guide for picking the national champion than the various polls.

"I would say so, yeah," he said, "because theirs is basically based on record, and that's pretty much it. Our guys, they rate out the defense, the offense, so really their numbers are more of an indicator of how good those teams are. And, of course, it's all done for betting purposes, but that's really the telltale sign: is a team three points better or is it three points worse?"

Curious as to what the current Oddsmaker Top 25 looks like? Well, here you go (AP rank in parentheses):

Ohio State (1)
Texas (6)
Southern Cal (3)
Michigan (4)
Florida (2)
Cal (10)
LSU (14)
Louisville (7)
Tennessee (8)
Notre Dame (9)
Clemson (12)
Oregon (18)
West Virginia (5)
Oklahoma (23)
Nebraska (21)
Auburn (11)
Wisconsin (25)
Missouri (19)
Boise State (20)
Georgia Tech (13)
Miami (NR)
Virginia Tech (23)
Iowa (15)
Penn State (NR)
BYU (NR)

Wanna see something else that'll get Horns fans blood churning? Jeff Sagarin's computer ratings have Texas at #2, just behind Ohio State. (This is in Jeff's "Predictor" ratings, which he says are his most accurate. Naturally, the BCS uses ELO, which can't factor in margin for victory.)

What's the bottom line? There are several points I want to put up for discussion.

Does the above Top 25 mean that Vegas would make Texas a favorite over everyone ranked below them, were the game played on a neutral field?
I ask this because lines are, in the end, set to create action. Vegas isn't just about picking the right spread - it's about creating action. And balance on the action. Doesn't Notre Dame have to lay more points because so many folks bet on the Irish? Is this a consideration in these rankings? Or is this just real, hardcore, do-it-for-a-living experts ranking their teams? These are the questions we have to ask.

But damn, this is interesting. Thanks, Dan.

--PB--

0 recs | Comment 12 comments

Story-email Email Printer Print

Comments

I'm going with no.

Not because they don't know what they're talking about, because in truth, they do seem to do a lot more research than any AP or coach, or Harris poll voter. As you say, it's their job. But also, it's their job to create lines. Not to be a predictor of who would win.

If the NCAA would just bite the bullet and pay a group of people to seriously research this, i.e. the Harris group, but more credible, then we would probably have a more accurate poll.

Hook 'em. 'Nuff said.

by GoHorns on Oct 12, 2006 12:50 PM CDT reply actions 0 recs

What about the Master Coaches Survey

They research it already. Just check out their website.

by aorist9 on Oct 12, 2006 1:39 PM CDT up reply actions 0 recs

Maybe, but not quite

The only problem with the Master Coaches is that they still have some ties to programs and may be a little skewed in their assessments. I would think that this group is probably the most credible of any poll out there, but it's still a little too close for me to truly believe in.

Hook 'em. 'Nuff said.

by GoHorns on Oct 12, 2006 2:08 PM CDT up reply actions 0 recs

Sagarin

So if you go by his Predictor, Florida beats Auburn this weekend, Cal beats USC and Louisville beats WVU and Rutgers.

On the Oddsmaker Top 25... do you really think this is what they predict as the outcome, or are their results skewed by the size of the respective fan bases? I mean, they want to drive business, so of course they would have Louisville, WVU, Rutgers and Boise State way down on the poll.

Real men DO watch volleyball

by patienthornsfan on Oct 12, 2006 1:45 PM CDT reply actions 0 recs

Uh, yeah...

I'm not sure having Cal three spots above Tennessee is "more accurate." Plus, two-loss Oklahoma at #14, two-loss Miami (who has struggled all year) at #21, and two-loss BYU at #25 make this poll about like all the others. There seem to be just as many flaws. It is interesting, though.

by imarealist on Oct 12, 2006 2:02 PM CDT reply actions 0 recs

And...

I completely overlooked two-loss LSU at #7, which may be the most ridiculous ranking of all.

by imarealist on Oct 12, 2006 2:03 PM CDT up reply actions 0 recs

Polls

I think, as much as everyone hates it, the data supports that the computers rankings WITH margin of vicotry included are the most accurate predictor of future success.

USC UTHSCSA

by txmed on Oct 12, 2006 2:08 PM CDT reply actions 0 recs

imasofullofmyself

first off no one said the Vegas rankings were perfect.

secondly, you can argue in support of their rankings all day, as it is really all hypothetical anyways and is not solely based on record (who plays on neutral fields besides TX/OU? - the cocktail party ain't neautral). As stated their rankings are based on if the teams line up tomorrow at a neautral site who would win, which I feel is a very fair way to rank.

In response to the realist:

the cal that is playing in week 6 could very well be beat UTenn on a neutral field.
OU really only has 1 loss, and it was to the #2 team. The could easily beat anyone from 15-25 on a neautral field.
BYU - who cares, and who knows.

Also, PB properly addressed the school's fans betting impact. However, I would argue that the lines are generally opened at what they believe the correct spread of the game to be, and then adjusted w/ juice cuts as well to create "action". I am continually amazed when bookies set a line at 35, and the kicker misses a PAT to only win by 34.

Anyways, if you follow CFB gambling, Vegas knows WTF they are doing, so prancing around and calling out individual rankings of theirs based on your opinion makes you sound full of yourself.

Crystal Balls

by MMHorns on Oct 12, 2006 2:20 PM CDT reply actions 0 recs

Post

Your post is excellent. The idea of asking odds makers to create a poll is intriguing, but I agree with several of the comments above that the end result is not more persuaive than the current rankings. Indeed, some of the odds maker rankings are bizarre . . . although on a sentimental level I like Texas at 2 and W.Va. at 13. One has to think the driver for their poll is action, not true relative positioning.

Thanks also to aorist9 for the link to the MCS, of which I was unaware.

Finally, I think USC is right with regard to Sagarin's Predictor rankings, although it creates the unfortunate incentive to blow out weak opponents instead of using them to develop talent otherwise sitting on the bench. Worse, it also kills sportsmanship, for example taking a knee on the last play instead of throwing a Hail Mary into the endzone with a 30 point lead (although it would have no effect on Spurrier's play calling).

by Allaha on Oct 12, 2006 2:31 PM CDT reply actions 0 recs

One note

The margin of victory stops after a certain level. In other words, once you're up by 35 or whatever, there's no additional reward for winning by more.

by PB @ BON on Oct 12, 2006 2:33 PM CDT up reply actions 0 recs

thanks

Thanks: I (and perhaps certain coaches) did not know this. Furthermore, that policy makes sense, given that I do not think a 38-3 victory is meaningfully different from a 63-3 victory -- unless, of course, it is against OU.

by Allaha on Oct 12, 2006 2:54 PM CDT up reply actions 0 recs

it's half way towards an answer.

just about anyone will have a slant on their analysis. the real problem is not enough data points. Most teams play at least half their schedule aginst teams that are pretty low on the competitive scale and that makes quite a bit of your analysis about just how well a team plays against a team with no chance of winning. This has little relation to how well a team plays against a equally competitive team. So you have even less data to work with. Plus you have injuries and personal problems of the players you never here about and many facets that control the out come may not make into an anaylsis procedure. So you have to make predictions based on any where from 4 to 8 proformances which is never going to be enough to get predictive projections.
it sucks to try to order teams for bowl games but this is not always a bad thing that means you have to watch the game to see who wins.

xerxes

by Xerxes on Oct 13, 2006 9:05 AM CDT reply actions 0 recs