I'm mad but I'm giving up

[jay santos]

Originally Posted by Ace Tomato Co.

OK, just to clarify: you are saying that to the computers SOS is irrelevant? Obviously you don't feel SOS is irrelevant in judging a team's actual strength?

SOS is irrelevant in the sense that in August if you field a team with players and preparation and coaching that equate to a #12 team in the nation and you play like a #12 team in the nation all year, with no surprises, that you will end up #12 in the computers. It doesn't matter what your SOS is. It matters how good you are and how you play.

If that team had an SOS of #120, then they go 12-0, if the SOS was tougher they go 10-2, if it's brutal they might go 7-5/8-4, but the team is the same team with the same strength and the computers will evaluate them the same no matter the SOS/record combination and put them at #12.

That is why in theory there is no way to trick the system. You're not going to go undefeated even with a #120 SOS, unless you really are good enough to be #12. And if you're good enough to be #12 and play like it all season, you'll end up #12 no matter the SOS/record combination.

[pelagius]

Originally Posted by jay santos

If that team had an SOS of #120, then they go 12-0, if the SOS was tougher they go 10-2, if it's brutal they might go 7-5/8-4, but the team is the same team with the same strength and the computers will evaluate them the same no matter the SOS/record combination and put them at #12.

That is why in theory there is no way to trick the system. You're not going to go undefeated even with a #120 SOS, unless you really are good enough to be #12. And if you're good enough to be #12 and play like it all season, you'll end up #12 no matter the SOS/record combination.

Jay, the one caveat to this is the standard error of the estimated ranking. It is pretty wide. In most models you will have a hard time statistically rejecting the hypothesis that the 12th ranked and 1st ranked team have same true rating (or the 12th compared to the 24th).

Also, parameter uncertainty is going to probably be the highest for undefeated teams with weak strengths of schedule.

But, other than that, I agree with you.

[Indy Coug]

If Hawaii was barely better than the competition they played against (let's say the probability of each win was 55%), the odds of them going 12-0 would be 1,305:1

If the a priori probability of Hawaii going undefeated was 50%, that would imply that the equal probability of winning each game was 94.4%. Doesn't that at least give SOME credence to Hawaii actually being a half-decent team?

I don't think anyone's foolish enough to think Hawaii is the best team in the country because they have the best record, but I think it's equally foolish to think they aren't any good (i.e. ranked 10th to 20th) because their schedule was ridiculous.

[jay santos]

Originally Posted by pelagius

Jay, the one caveat to this is the standard error of the estimated ranking. It is pretty wide. In most models you will have a hard time statistically rejecting the hypothesis that the 12th ranked and 1st ranked team have same true rating (or the 12th compared to the 24th).

Also, parameter uncertainty is going to probably be the highest for undefeated teams with weak strengths of schedule.

But, other than that, I agree with you.

Yes, I'm talking at a very theoretical level, that's why I add "if a team plays exactly like the 12th best team all year" eliminating the variance issue.

[pelagius]

Originally Posted by jay santos

Yes, I'm talking at a very theoretical level, that's why I add "if a team plays exactly like the 12th best team all year" eliminating the variance issue.

I shouldn't quibble with the your statement because the intent of your clarification was to make clear that you were abstracting from the estimation error. However, I am going to quibble anyways. In order to get the estimation error to go to zero you need a stronger statement: "the "ith" best team plays like the "ith" best team all year for all "i". Ok, now I feel much better and agree with your statement without reservation.

[jay santos]

Originally Posted by pelagius

I shouldn't quibble with the your statement because the intent of your clarification was to make clear that you were abstracting from the estimation error. However, I am going to quibble anyways. In order to get the estimation error to go to zero you need a stronger statement: "the "ith" best team plays like the "ith" best team all year for all "i". Ok, now I feel much better and agree with your statement without reservation.

the ith best team plays like the ith best team on every down of every game. and so does everybody else.

[BYU71]

Who the hell cares about strenght of schedules and how crappy a schedule can you get to get to the BCS ball game.

So are you telling me with BYU all they care about is getting to a BCS bowl game and getting the money. Are you telling me they will provide Chuckarama type meals for years and years for a shot at getting the money.

Man, I thought BYU was a special place. It is just like any other place, SHOW ME THE MONEY!!!!

This is probably OK with me. If they are going to screw the fans with lousy games at LES to get the money, they will probably turn their heads on HC violations to get there too.