College Gameday on undefeated teams

[Dossenator]

Did anyone see College Gameday yesterday? They showed the list of all the undefeated teams before yesterdays games. They said they ran each of their schedules through the computer 10,000 times to determine the chance each of them had of going undefeated.

Very interesting numbers came out:

Boise had almost a 100% chance of going undefeated (from memory it was 98 point something percent).

The next closest was barely over 40% chance.

Then Auburn and LSU came in at less than 2% chance of going undefeated....Oregon and Missouri were in the same boat.

I can't find the list but found it interesting when they showed it yesterday.

This clearly illustrates how Boise's schedule will probably get them into the big game when all the undefeated teams go down at least one time....I don't think there will be more than 1 team from a AQ conference go undefeated....maybe 0.

[donovan]

Did anyone see College Gameday yesterday? They showed the list of all the undefeated teams before yesterdays games. They said they ran each of their schedules through the computer 10,000 times to determine the chance each of them had of going undefeated.

Very interesting numbers came out:

Boise had almost a 100% chance of going undefeated (from memory it was 98 point something percent).

The next closest was barely over 40% chance.

Then Auburn and LSU came in at less than 2% chance of going undefeated....Oregon and Missouri were in the same boat.

I can't find the list but found it interesting when they showed it yesterday.

This clearly illustrates how Boise's schedule will probably get them into the big game when all the undefeated teams go down at least one time....I don't think there will be more than 1 team from a AQ conference go undefeated....maybe 0.

Because some self-serving putz's on ESPN say so.....and why do you have to run a program worth its salt, 10,000 times unless you want to try and dispel the earn reputation of intellectual pygmies.

[Dossenator]

Did anyone see College Gameday yesterday? They showed the list of all the undefeated teams before yesterdays games. They said they ran each of their schedules through the computer 10,000 times to determine the chance each of them had of going undefeated.

Very interesting numbers came out:

Boise had almost a 100% chance of going undefeated (from memory it was 98 point something percent).

The next closest was barely over 40% chance.

Then Auburn and LSU came in at less than 2% chance of going undefeated....Oregon and Missouri were in the same boat.

I can't find the list but found it interesting when they showed it yesterday.

This clearly illustrates how Boise's schedule will probably get them into the big game when all the undefeated teams go down at least one time....I don't think there will be more than 1 team from a AQ conference go undefeated....maybe 0.

Because some self-serving putz's on ESPN say so.....and why do you have to run a program worth its salt, 10,000 times unless you want to try and dispel the earn reputation of intellectual pygmies.

10,000 times for margin of error. Same as a political poll...why call 10,000 people? Should we just ask the bum on the street corner who he voted for and when he gives his answer declare that candidate the winner.

[WoVeU]

I saw the show and the numbers. The Big 12 teams had really small chances.

But what do they mean by the computers...a program obviously...but what program? Most comparisons and calculations I have seen from these folks rely on SOS and similar inptus, these do not require progression or iteration...it comes down to probabilities and weightings. You don't need 10,000...you need 1 to "estimate" a probability of going undefeated.

To run something that many times you are amassing a distribution, a discrete cumulative distribution function, etc. But this would be more in line with estimating poll progressions given the likelihood of all the teams winning or losing from week to week. But I think they throw around a number like that because only 1 in 10,000 viewers would have a clue about the validity of such a premise!

[Brian Roastbeef]

Sounds like what Nate Silver does for politics. He runs his polls through some computer setup 100,000 times or so, then tries to work up an argument as to why he believes Democrats won't get pounded quite as badly as his numbers tell him they will....

Anyway, lets just look at the schedules for the remaining undefeateds ourselves...

Oregon: at USC, Washington, at California, Arizona, at Oregon State
Boise St. Louisiana Tech, Hawaii, at Idaho, Fresno State, at Nevada, Utah State
TCU at UNLV, at Utah, San Diego St, New Mexico
Utah at Air Force, TCU, at Notre Dame, at San Diego St, BYU
Michigan St. at Iowa, Minnesota, Purdue, Penn St.
Missouri at Nebraska, at Texas Tech, Kansas St, at Iowa St, Kansas (Big XII Title Game - could be just about anybody - Oklahoma again, OSU, or even Baylor...)
Auburn at Ole Miss, Chattanooga, Georgia, at Alabama (SEC Title Game - right now looks like South Carolina)


First thing, obviously is that either TCU or Utah will lose.

Beyond that, I don't see any schedule where I could say that anybody is guaranteed to win out, not even Boise St. They catch a break in getting Hawaii on the Smurf Turf. Unless Boise's defense somehow collapses and it turns to a shootout, then I think Boise wins that easily... Fresno St. and Nevada are not pushovers though. Fresno they get at home, but Nevada is on the road. Boise probably still wins out, but I think collectively those teams are worth maybe 7-8%... If they had to play those remaining teams 12 times each, finishing 71-1 wouldn't be unlikely, right...?

TCU and Utah, have to contend with each other. Beyond that, both have somewhat of an upset watch in San Diego St. Utah has Notre Dame, who could provide problems, if Utah hasn't improved from the team that needed overtime to beat Pitt. Air Force has been okay this year, but can't go the distance against tougher competition. I say TCU's chances are roughly whatever their money line against Utah is. Utah is somewhat more of a longshot, both due to being the likely underdog vs. TCU and because of ND.

Michigan St. faces their most difficult test next week. Wisconsin barely handled Iowa. Michigan couldn't. And everybody was just waiting around for the Hawkeyes to screw up Ohio St's season before Wisconsin beat them to it. Beyond that they have a very weak Minnesota, a weak Purdue that managed an upset against Northwestern, and a very disappointing Penn St. I say MSU's chances are only slightly less than their chances of beating Iowa. Figure that they are a slight favorite at Iowa, maybe 55%? Then their chances of going undefeated are about 50%.

Missouri has the nastiest remaining schedule of the undefeateds in my opinion. I can see them favored in all of these games, but not terribly much so. Nebraska is a test. Kansas St. is a test. Texas Tech and Iowa St are dangerous chances to be upset, as the latter proved against Texas this week. If they get through all of that then they have the title game... which means there is a good chance that Oklahoma comes back to greet them 2007-style. I'd have to do some guessing and math to figure Missouri's chances, but to give a rough guess... 10-15%?

Auburn has a couple of stops on their schedule too. Mississippi will probably give more of a fight than Georgia this year. Alabama is the obvious potential spoiler. I'd guess they have a 90% chance of getting to Alabama undefeated. Reduce that by their chance of beating Alabama, and then they get to play in the SEC title game... That will probably be South Carolina or Florida... If you assume 50/50 vs. Alabama, and that they would be a reasonable favorite to beat SC or Fla. Then roughly 25-33% for them.

[WoVeU]

I agree with all of that (give or take maybe a few %) except for Auburn. Georgia is bringing it right now, if I was Auburn I would be wishing we had gotten them at the opening of the conference schedule. Georgia appears to be rolling right now! I'd put Auburn at about a 10-12.5% chance of making it. Similar to Mizz.

[donovan]

I saw the show and the numbers. The Big 12 teams had really small chances.

But what do they mean by the computers...a program obviously...but what program? Most comparisons and calculations I have seen from these folks rely on SOS and similar inptus, these do not require progression or iteration...it comes down to probabilities and weightings. You don't need 10,000...you need 1 to "estimate" a probability of going undefeated.

To run something that many times you are amassing a distribution, a discrete cumulative distribution function, etc. But this would be more in line with estimating poll progressions given the likelihood of all the teams winning or losing from week to week. But I think they throw around a number like that because only 1 in 10,000 viewers would have a clue about the validity of such a premise!

Exactly!!

[Brian Roastbeef]

I agree with all of that (give or take maybe a few %) except for Auburn. Georgia is bringing it right now, if I was Auburn I would be wishing we had gotten them at the opening of the conference schedule. Georgia appears to be rolling right now! I'd put Auburn at about a 10-12.5% chance of making it. Similar to Mizz.

Could be... so far I'm thinking Georgia's recent success has been due to the weakness of Tennessee and Vandy, though admittedly beating Kentucky was more of a victory than their record would suggest.

If Georgia hangs with Florida - keeps them tied at the half, beats the spread, something like that - then I'll probably join those who are giving that Georgia-Auburn date in November a close look, particularly Auburn could be thinking about their week off, and then their visit to Tuscaloosa.

[WoVeU]

I saw the show and the numbers. The Big 12 teams had really small chances.

But what do they mean by the computers...a program obviously...but what program? Most comparisons and calculations I have seen from these folks rely on SOS and similar inptus, these do not require progression or iteration...it comes down to probabilities and weightings. You don't need 10,000...you need 1 to "estimate" a probability of going undefeated.

To run something that many times you are amassing a distribution, a discrete cumulative distribution function, etc. But this would be more in line with estimating poll progressions given the likelihood of all the teams winning or losing from week to week. But I think they throw around a number like that because only 1 in 10,000 viewers would have a clue about the validity of such a premise!

Exactly!!

There is a reason they call it media. And they are doing all of the mediating they can. Just because they are sports-based this doesn't make them any different than any other network, ABC, FOX, NBC, and the rest. They do report "news" because it is news to people. But they don't stop there, everyone always has an agenda...really agendas. People's politics and their portfolio will always bubble up and produce answers to questions like: "How do we do this or that?" "What else should we do?"
They all like to interpret the "news" not just report it! People begin being concerned that others may come to a different conclusion before they even finish stating the evidence.
As you say, "First the hanging (fiery stakes, the chair) then the trial!" Evidence is always more palpable in the excusatory mode as compared to the exculpatory!

[billybud]

OK...I'll put it to the test to any taker.

PM me. I'll take Boise to undefeated in regular season for $250.

[WoVeU]

I'd bet $1000 and give 6 points on all home games on Boise State going undefeated.

[silverfox]

Vegas is waiting.....

You can design desired outcomes in a computer program.

[..fanatic]

Vegas is waiting.....

You can design desired outcomes in a computer program.

I feel so enlightened. Any other thoughts of masterful wisdom?

[donovan]

I thought that Boise State and TCU played each other during the regular season also. So that is one loss for one of them.

I think they play on the 12th of November in Boise.