College Football - Week of Sept. 19th

[Guest]

I like the CBSSports 120 poll in concept because it actually attempts to rank from 1-120, rather than support the idea that 25 teams are worth talking about at any given time. That said, I don't get this week's poll one bit.

 

http://www.cbssports.com/collegefootball/polls/120

 

It's good for poop and amusement, I guess.

 

If there are 120 teams in FBS and that's what they're ranking, how is Florida's SOS 121?

their opponents have been that bad

 

Can't really argue with that one.

[snoodmonger]

I'm wondering if Renfree will get the start over Lewis for Duke

 

We should destroy Duke, but we've been kind of sloppy so far this year. I think KU is favored by 20 points.

[Guest]

This week is a big week for Notre Dame. If they recover from a ridiculously tough loss, get their stuff together, play like the veterans they are, and pound MSU, things look a lot better. If not, things could unravel very, very quickly.

 

As for Ball State, if they don't beat Army, they may not win a game.

MSU lost to CMU. Although Central is probably the best team in the MAC, that ain't saying much.

 

I think ND will crush them.

[SouthSideRyan]

This week is a big week for Notre Dame. If they recover from a ridiculously tough loss, get their stuff together, play like the veterans they are, and pound MSU, things look a lot better. If not, things could unravel very, very quickly.

 

As for Ball State, if they don't beat Army, they may not win a game.

MSU lost to CMU. Although Central is probably the best team in the MAC, that ain't saying much.

 

I think ND will crush them.

 

And that's saying something.

[Guest]

This week is a big week for Notre Dame. If they recover from a ridiculously tough loss, get their stuff together, play like the veterans they are, and pound MSU, things look a lot better. If not, things could unravel very, very quickly.

 

As for Ball State, if they don't beat Army, they may not win a game.

MSU lost to CMU. Although Central is probably the best team in the MAC, that ain't saying much.

 

I think ND will crush them.

 

And that's saying something.

I don't have to like it.

 

Do I?

[Warpticon]

Am I crazy for feeling really good about Arkansas's chances against Georgia? Them giving up 37 to SC last week pretty much erased every ounce of intimidation I had.

[Andy]

This week is a big week for Notre Dame. If they recover from a ridiculously tough loss, get their stuff together, play like the veterans they are, and pound MSU, things look a lot better. If not, things could unravel very, very quickly.

 

As for Ball State, if they don't beat Army, they may not win a game.

MSU lost to CMU. Although Central is probably the best team in the MAC, that ain't saying much.

 

I think ND will crush them.

If there is one thing Michigan State has proven beyond reasonable doubt, it's that their performances against other teams have nothing to do with their performance against Notre Dame. Hell, in 2005 they were terrible, but played like world-beaters in South Bend. In 2006 they weren't much better but played like world-beaters for two and a half quarters.

 

I fully expect they will play well above their heads like usual.

 

(ND should still beat them. But they 'should' beat them a lot of times.)

[bukie]

Because I hate the human polls at this point, here's the top 25 according to Sagarin, two ways:

 

1) Pure Points (where scoring margin is all that matters - also the best predictor of future results, in general):

 

1. USC

2. Oklahoma

3. BYU

4. Ohio State

5. Texas

6. Boise State

7. LSU

8. California

9. Penn State

10. Alabama

11. Florida

12. Virginia Tech

13. Nebraska

14. Notre Dame

15. TCU

16. Tennessee

17. Missouri

18. Iowa

19. Michigan

20. Clemson

21. South Carolina

22. Arizona State

23. Oklahoma State

24. Auburn

25. Oregon

 

2) Elo-Chess (where only winning and losing matters...i.e. the BCS version):

 

1. USC

2. Boise State

3. Alabama

4. Ohio State

5. LSU

6. BYU

7. Utah

8. Miami-FL

9. California

10. Oregon

11. Houston

12. North Carolina

13. Missouri

14. TCU

15. Oklahoma State

16. Georgia

17. Virginia Tech

18. Georgia Tech

19. Texas

20. Florida

21. Wisconsin

22. Arizona

23. UCLA

24. Michigan

25. Baylor

[Grit amp Heart]

thats worse.

[bukie]

thats worse.

I disagree. At least that's actually based on the results on the field, and not some imaginary level of competency assigned to a team in the preseason.

[Andy]

thats worse.

I disagree. At least that's actually based on the results on the field, and not some imaginary level of competency assigned to a team in the preseason.

If on-field results lead you to believe that Virginia Tech or Missouri are top 20 teams right now, then your system is wrong. If on-field results lead to Miami being a top 10 team, the system is wrong.

[jersey cubs fan]

Because I hate the human polls at this point, here's the top 25 according to Sagarin, two ways:

 

1) Pure Points (where scoring margin is all that matters - also the best predictor of future results, in general):

 

1. USC

2. Oklahoma

3. BYU

4. Ohio State

 

Ohio State won by 4 and lost by 3, if scoring margin is all that matters how do they rank 4?

[bukie]

thats worse.

I disagree. At least that's actually based on the results on the field, and not some imaginary level of competency assigned to a team in the preseason.

If on-field results lead you to believe that Virginia Tech or Missouri are top 20 teams right now, then your system is wrong. If on-field results lead to Miami being a top 10 team, the system is wrong.

No, it just means the system disagrees with common perception. Except with Virginia Tech, which is a top 20 team everywhere except apparently in your mind.

 

The points system is by far the more accurate of the two, as the inability to take margin into effect severely limits the accuracy of any analysis.

[bukie]

Because I hate the human polls at this point, here's the top 25 according to Sagarin, two ways:

 

1) Pure Points (where scoring margin is all that matters - also the best predictor of future results, in general):

 

1. USC

2. Oklahoma

3. BYU

4. Ohio State

 

Ohio State won by 4 and lost by 3, if scoring margin is all that matters how do they rank 4?

By taking into account opponents scoring margin and opponents' opponents scoring margin. Their strength of schedule is measured higher than any other top 50 team at this point by playing two quality teams to this point.

[Grit amp Heart]

thats worse.

I disagree. At least that's actually based on the results on the field, and not some imaginary level of competency assigned to a team in the preseason.

 

For the first few weeks of the season, the starting ratings have weight

in the process

 

http://www.usatoday.com/sports/sagarin/fbt09.htm

 

Maybe by the end of the season they will be better, but not right now. Alabama beat Virginia Tech by 10 points and then beat FIU #113, by 26 points. Oklahoma lost to BYU by 1 point and then beat #151 by 64 points. So OU's win over a crappy team by 64 points outweighs Alabama beating the #12 team by 10 points enough to cancel out the loss to BYU?

[bukie]

thats worse.

I disagree. At least that's actually based on the results on the field, and not some imaginary level of competency assigned to a team in the preseason.

 

For the first few weeks of the season, the starting ratings have weight

in the process

 

http://www.usatoday.com/sports/sagarin/fbt09.htm

 

Maybe by the end of the season they will be better, but not right now. Alabama beat Virginia Tech by 10 points and then beat FIU #113, by 26 points. Oklahoma lost to BYU by 1 point and then beat #151 by 64 points. So OU's win over a crappy team by 64 points outweighs Alabama beating the #12 team by 10 points enough to cancel out the loss to BYU?

Since BYU is a better team than Virginia Tech, yes. It's not perfect by any means, but it's a better measure than human polls.

[Grit amp Heart]

and why is BYU a better team than Virginia Tech? Because they beat #118 Tulane by 51 and VT beat #117 Marshall by 42? It's not because BYU beat Oklahoma, because then you're in a circular argument.

 

That's enough to make Oklahoma's loss to BYU better than Alabama's win over Virginia Tech?

[TruffleShuffle]

This week is a big week for Notre Dame. If they recover from a ridiculously tough loss, get their stuff together, play like the veterans they are, and pound MSU, things look a lot better. If not, things could unravel very, very quickly.

 

As for Ball State, if they don't beat Army, they may not win a game.

MSU lost to CMU. Although Central is probably the best team in the MAC, that ain't saying much.

 

I think ND will crush them.

 

you can't really judge mich st like that, though. they'll do things like getting blown out by northwestern and then playing ohio st tight in the horseshoe. they're always schizo.

[bukie]

and why is BYU a better team than Virginia Tech? Because they beat #118 Tulane by 51 and VT beat #117 Marshall by 42? It's not because BYU beat Oklahoma, because then you're in a circular argument.

 

That's enough to make Oklahoma's loss to BYU better than Alabama's win over Virginia Tech?

 

Why is BYU a better team than Virginia Tech? Because they performed better in two wins versus comparable opponents (SOS 34 vs. 39) than VT did in a win and a loss.

 

Why is Oklahoma rated better than Alabama? Because they performed better in a win and a loss versus somewhat comparable opponents (SOS 60 vs. 47) than Alabama did in two wins. And yes, winning by 50 makes more of a difference than winning by 23, even with diminishing returns awarded for higher margins, as Sagarin uses.

 

There is currently a very small sample size, and not every team is connected yet (meaning there is a way to link every team to every other team through games played), so there are obviously going to be more flaws in the ratings early than as the season progresses. I'm sure the system also doesn't take into account that, for example, Oklahoma's QB is out the first half of the year, so those results will be affected as such.

 

At this point, even one week's worth of results adds an additional 50%+ of results to the current set of data (since there are at most 2 weeks of results for teams at this point), so the ratings will continue to fluctuate wildly from week to week, even to the end of the season. To me, though, even with its limitations, it's still a better gauge of current results than the human polls, which rarely take actual results into the voting.

[Grit amp Heart]

Okay if everything is determined by points and wins and losses, how are all the Ivy League schools rated differently? Shouldn't they be tied? None of them have a played a single game, yet Harvard is #120 and Columbia is #212. If the rankings are determined by what teams have done this year on the field, how the hell can Harvard be 92 spots ahead of Columbia.

 

Does it include practices

[Andy]

I don't think Va Tech should be a top 20 team in any poll for the record. That's why I said that - they were dominated by Bama even if the score didn't reflect it.

 

I don't understand how Miami could possibly be ranked 8th when their only game so far was a 4-point win over a team that should've lost to freaking Jacksonville State. There is no way to justify that.

[jersey cubs fan]

Okay if everything is determined by points and wins and losses, how are all the Ivy League schools rated differently? Shouldn't they be tied? None of them have a played a single game, yet Harvard is #120 and Columbia is #212. If the rankings are determined by what teams have done this year on the field, how the hell can Harvard be 92 spots ahead of Columbia.

 

Does it include practices

 

Columbia sucks.

[bukie]

Okay if everything is determined by points and wins and losses, how are all the Ivy League schools rated differently? Shouldn't they be tied? None of them have a played a single game, yet Harvard is #120 and Columbia is #212. If the rankings are determined by what teams have done this year on the field, how the hell can Harvard be 92 spots ahead of Columbia.

 

Does it include practices

Because until all teams are connected, he makes use of results prior to this year to connect the teams (i.e. the Bayesian process). Thus, ratings are generated from past results to connect teams that are not yet connected from current results.

[bukie]

I don't think Va Tech should be a top 20 team in any poll for the record. That's why I said that - they were dominated by Bama even if the score didn't reflect it.

 

I don't understand how Miami could possibly be ranked 8th when their only game so far was a 4-point win over a team that should've lost to freaking Jacksonville State. There is no way to justify that.

The Elo rating is based solely on whether a team wins or loses. Therefore, it's more a measure on who has the best wins. Miami has only one win as a sample size over what is ostensibly a top-40 team, which at this point in the season is better than the vast majority of other teams' best wins.

[shnsajax]

I think those polls are just as meaningless after two weeks as any other polls. The only thing you can do is watch the games and try to make the most subjective decision based on how those teams looked and the players that they have. Human polls don't do a great job at this, but after two weeks computers do an even worse job. By the end of the year once the computers have enough data to go off of then they can make a more accurate assessment, but not today.