cartman's FOFC College Football Top 25, Week 11 open for voting

[cartman]

Time to open up the voting for this week. Submissions are due by kickoff of the North Carolina-Virginia Tech game on Thursday, Nov. 17th. Here is how last week's poll fared:

Team Rec. Weekend Results
1. LSU 10-0 beat Western Kentucky 42-9
2. Stanford 9-1 lost to #6 Oregon 53-30
3. Oklahoma State 10-0 beat Texas Tech 66-6
4. Boise State 8-1 lost to #25 TCU 36-35
5. Alabama 9-1 beat Mississippi State 24-7
T6. Oregon 9-1 beat #2 Stanford 53-30
T6. Oklahoma 8-1 Bye
8. Clemson 9-1 beat Wake Forest 31-28
9. Virginia Tech 9-1 beat #24 Georgia Tech 37-26
10. Arkansas 9-1 beat Tennessee 49-7
11. Houston 10-0 beat Tulane 73-17
12. Penn State 8-2 lost to #19 Nebraska 17-14
13. Wisconsin 8-2 beat Minnesota 42-13
14. South Carolina 8-2 beat Florida 17-12
15. Georgia 8-2 beat Auburn 45-7
16. Michigan State 8-2 beat Iowa 37-21
T17. USC 8-2 beat Washington 40-17
T17. Cincinnati 7-2 lost to West Virginia 24-21
T19. Southern Miss 9-1 beat Central Florida 30-29
T19. Nebraska 8-2 beat #12 Penn State 17-14
21. Michigan 8-2 beat Illinois 31-14
22. Texas 6-3 lost to Missouri 17-5
23. Kansas State 8-2 beat Texas A&M 53-50 (4OT)
24. Georgia Tech 7-3 lost to #9 Virginia Tech 37-26
25. TCU 8-2 beat #4 Boise State 36-35

[duckman]

Louisiana State
Oklahoma State
Oregon
Oklahoma
Alabama
Arkansas
Clemson
Virginia Tech
Stanford
Houston
South Carolina
Boise State
Kansas State
Nebraska
Georgia
Michigan State
Wisconsin
Texas Christian
Penn State
West Virginia
Southern California
Michigan
Southern Mississippi
Baylor
Notre Dame

[jbergey22]

1. LSU
2. Oklahoma State
3. Alabama
4. Oklahoma
5. Oregon
6. Arkansas
7. Stanford
8. Wisconsin
9. Clemson
10. Virginia Tech
11. Nebraska
12. S Carolina
13. Boise St
14. Georgia
15. Kansas St
16. Houston
17. USC
18. Michigan St
19. Southern Miss
20. TCU
21. Penn St
22. Michigan
23. Notre Dame
24. W Virginia
25. Florida St

[Cuckoo]

1. LSU (1)
2. Oklahoma State (3)
3. Oregon (6)
4. Alabama (4)
5. Oklahoma (7)
6. Arkansas (8)
7. Clemson (9)
8. Virginia Tech (10)
9. Stanford (2)
10. Houston (11)
11. Boise State (5)
12. Michigan State (13)
13. Wisconsin (14)
14. South Carolina (15)
15. Georgia (16)
16. Nebraska (17)
17. Kansas State (19)
18. Michigan (20)
19. USC (21)
20. Penn State (12)
21. TCU (24)
22. Southern Mississippi (25)
23. Florida State (NR)
24. Georgia Tech (18)
25. Notre Dame (NR)

With last week's ranking. Dropped out: Cincinnati, Texas

[britrock88]

1. Alabama
2. Louisiana State
3. Boise State
4. Oklahoma
5. Oklahoma State
6. Oregon
7. Michigan
8. Stanford
9. Wisconsin
10. Notre Dame
11. Texas Christian
12. Nebraska
13. Virginia Tech
14. Southern Mississippi
15. Southern California
16. Clemson
17. Florida State
18. Texas A&M
19. Georgia
20. Houston
21. Arkansas
22. South Carolina
23. Penn State
24. Tulsa
25. Toledo

[MJ4H]

1. Alabama
2. Louisiana State


rofl

[britrock88]

rofl

Heh, yeah.

[JonInMiddleGA]

1. LSU
2. Oklahoma State
3. Alabama
4. Oregon
5. Oklahoma
6. Arkansas
7. Virginia Tech
8. Houston
9. Stanford
10. Clemson
11. Michigan State
12. Wisconsin
13. TCU
14. Boise State
15. South Carolina
16. Nebraska
17. Georgia
18. USC
19. Kansas State
20. Michigan
21. Penn State
22. Southern Miss
23. West Virginia
24. Florida State
25. Tulsa

[jbergey22]

Heh, yeah.

Id actually rofl on Arkansas being #21 in your poll over the Bama #1.

[MJ4H]

Id actually rofl on Arkansas being #21 in your poll over the Bama #1.

This is the first week Arkansas has made his top 25 at all, so I gave that a pass.

[britrock88]

Id actually rofl on Arkansas being #21 in your poll over the Bama #1.

The math is weighing Arkansas down because they played a IAA team and New Mexico, which is a IAA team masquerading as a IA team. And although they crushed both of those schools, MOV is capped in this ranking, so Arkansas is deprived of that benefit. (By comparison, in the unbounded SRS, they're 16th.)

With only the one loss, though, it's certainly a valid point. But I feel like this ranking is okay because presuming Arkansas is beaten by LSU in a couple weeks, they won't automatically drop several spots as would be the case in most polls; in fact, LSU's impact in the strength of schedule might offset the margin of defeat altogether.

[Young Drachma]

1 Oklahoma State
2 LSU
3 Alabama
4 Stanford
5 Oklahoma
6 Houston
7 USC
8 Clemson
9 Arkansas
10 Boise State
11 Oregon
12 Virginia Tech
13 Michigan
14 South Carolina
15 Kansas State
16 Nebraska
17 Notre Dame
18 Tulsa
19 Penn State
20 Southern Mississippi
21 Wisconsin
22 Georgia
23 TCU
24 Texas
25 Michigan State

Top 120 (rankings-)

[MJ4H]

I'd really not like to get into this again, but it's definitely a valid point when teams Arkansas has beaten and with worse records are ranked in the top 12 for several weeks while Arkansas is unranked. I mean, your polls are completely indefensible from top to bottom. Alabama-1 and LSU-2 is just one of a huge list of examples of this.

Seriously, why even play the games? The games themselves >>>>> the math.

[Galaxy]

1) LSU
2) Oklahoma State
3) Oregon
4) Alabama
5) Oklahoma
6) Arkansas
7) Virginia Tech
8) Stanford
9) Clemson
10) Houston
11) Wisconsin
12) Boise State
13) Kansas State
14) Michigan State
15) South Carolina
16) Nebraska
17) Michigan
18) Georgia
19) Penn State
20) USC
21) Southern Mississippi
22) Baylor
23) TCU
24) Cincinnati
25) Notre Dame

[Galaxy]

This is the first week Arkansas has made his top 25 at all, so I gave that a pass.

LOL. Can you imagine if that was used in real life?

[tarcone]

1. LSU
2. Oklahoma St.
3. Oklahoma
4. Oregon
5. Alabama
6. Arkansas
7. Clemson
8. Virginia Tech
9. Stanford
10. Houston
11. Wisconsin
12. South Carolina
13. Georgia
14. Michigan St.
15. Kansas St.
16. Boise St.
17. Nebraska
18. USC
19. Michigan
20. TCU
21. Southern Mississippi
22. West Virginia
23. Florida St.
24. Penn St.
25. Cincinnati

[Suicane75]

1. LSU
2. Oklahoma State
3. Alabama
4. Oregon
5. Oklahoma
6. Houston
7. Arkansas
8. Clemson
9. Virginia Tech
10. Stanford
11. USC
12. Georgia
13. Boise State
14. Wisconsin
15. Michigan State
16. South Carolina
17. Michigan
18. Southern Miss
19. TCU
20. Nebraska
21. Cincinnati
22. Kansas State
23. Texas
24. Notre Dame
25. Penn State

[britrock88]

Annar Rankings - College Football Computer Rankings (http://www.annarrankings.com/)

See #39. :cool:

[Tigercat]

Wow, rankings that rank teams solely based on yardage, and not just yardage per game, but strength of win and schedule based on yardage. What a horrible, horrible idea. You play to win the game. Hello? You play to win the game. You don't play it to just get yards.

[MJ4H]

Wow, rankings that rank teams solely based on yardage, and not just yardage per game, but strength of win and schedule based on yardage. What a horrible, horrible idea. You play to win the game. Hello? You play to win the game. You don't play it to just get yards.

Exactly I don't see the point in ranking teams for anything other than the entire object of the game. Winning. Nothing else really matters. Citing other polls that are similarly flawed only proves that there is more than one person missing the point entirely out there.

[britrock88]

Exactly I don't see the point in ranking teams for anything other than the entire object of the game. Winning. Nothing else really matters. Citing other polls that are similarly flawed only proves that there is more than one person missing the point entirely out there.

So most polls implicitly combine W-L record and SoS; my ranking (and the simple rating score it's derived from) looks at margin of victory and SoS as measured by MOV; and Annar looks at yards per play while accounting for SoS through opponents' defenses' yards per play.

Winning is certainly the ultimate objective of the game, but there are discrete parts of the game that have goals of their own. MOV judges teams based on their success per drive, so to speak, by paying attention to the number of points scored and allowed. Within the drive, Annar looks at the successes of a team on a play-by-play basis.

I don't think that it's wrong to analyze teams based on any of these criteria; in fact, I'm sure there are many other inputs that could be informative as to the relative performance or strength of teams. But there's an interesting point raised in critiques of systems that look at factors other than wins and losses. Many of these critiques focus on how these rankings ignore an important part of the game, that is, the result in terms of winner and loser. That is valid, but shouldn't the reverse be true? Could others critique the number of polls that are subjective and correlate very strongly to W-L record because those polls perhaps underweight other measures of success such as margin of victory, yardage, turnover ratio, etc.?

In an effort to clarify why polls that use data other than wins and losses exist, I'll just pull a couple points from the rankings I have. Alabama and LSU are 1 and 2, even though we are just weeks removed from LSU defeating Alabama 9 to 6 in overtime. To my spreadsheet, Alabama is not categorically ranked lower than LSU because of that loss. I see it that the fact that Alabama scored 3 fewer points than LSU is more informative than the actual victor/vanquished categorization, particularly because the game went to overtime. For another case, we'll take once-unbeaten Kansas State, who never cracked the Top 25 of my poll. I do not want to try to claim that this ranking has any predictive power. But the ranking did take into account the fact that Kansas State beat an FCS team (Eastern Kentucky) by only 3 points. (That is a _miserable_ outing, despite the tallied win.) That played no small part in holding Kansas State in the 30s, even as they won several games to start the season.

In short, I think rankings should work like statistical regressions, by factoring in more and more variables until much of the distribution of team performance is explained, then stripping away the least useful variables to produce a lean, but powerful measure for teams' effectiveness. I am just a guy with a spreadsheet and the ESPN scores page in front of me; I decided that MOV was the most informative and efficient category of data to rely on. Despite whatever irregularities the ranking may produce in comparison to subjective polls, I rest assured in that no cognitive biases color my perception of teams and affect the way my ranking plays out.

[One last point: Sagarin's ELO ranking is what the BCS uses (since it bars the use of margin of victory in its computer rankings (to keep down the Boise States of the world???)), but Sagarin also publishes his PREDICTOR ranking, which takes account of margin of victory, and directs readers to use that ranking for predictive purposes instead of the ELO ranking. Why? PREDICTOR has more data to better inform the results of the ranking!)

[MJ4H]

[One last point: Sagarin's ELO ranking is what the BCS uses (since it bars the use of margin of victory in its computer rankings (to keep down the Boise States of the world???)), but Sagarin also publishes his PREDICTOR ranking, which takes account of margin of victory, and directs readers to use that ranking for predictive purposes instead of the ELO ranking. Why? PREDICTOR has more data to better inform the results of the ranking!)

As I said, no amount of justifying things "logically" can get around the fact that you have teams that lost to other teams and with worse records ranked ahead of the other team. That is just not excusable.

And the reason the ELO rankings are used is because they use win/loss data only! Just like in chess (where Elo ratings are most commonly used), all that matters is who wins. The predictor rankings use all that other nonsense and come up with poorer rankings.

The predictor rankings (according to Sagarin) are better at guessing who would win in a matchup, but (and remember this next part), that isn't how teams should be ranked. We aren't asking "who is the best team" when we rank teams, we are asking "who has had the best season, according to the results of their games." These are not equivalent questions. Asking the wrong question leads to absurd answers.

Edit to add: Allowing nonsensical polls such as this is precisely why I don't participate in this poll. It's a complete waste of time.

[britrock88]

As I said, no amount of justifying things "logically" can get around the fact that you have teams that lost to other teams and with worse records ranked ahead of the other team. That is just not excusable.

There's an internal logic to your observation and your reaction to it, too. The syllogism looks something like this:

Givens:
1. In general, a team X that has lost to a team Y should not be ranked higher than team Y.
2. In general, a team W that has more losses than a team Z should not be ranked higher than team Z.
3. If it is the case that a team A has lost to a team B AND team A has more losses than team B, then team A absolutely should not be ranked higher than team B.
4. Violating premiss 1, 2, or 3 is inexcusable.

Assumptions:
1. Texas A&M has lost to Arkansas.
2. Texas A&M has lost more games than Arkansas.
3. Thus, Texas A&M has lost to Arkansas AND has lost more games than Arkansas.
4. Texas A&M is ranked higher than Arkansas.

Assumptions 3 and 4 violate Given 3, leading to the conclusion:
Ranking Texas A&M over Arkansas is inexcusable.

You have an internal logic which I respect. I only use a different approach to assessing teams' success, one that I would say takes into account many more contextual factors (relative strengths of schedule, convincingness of victories).


And the reason the ELO rankings are used is because they use win/loss data only!

The BCS's fiat is that it only allows the participation of computer rankings that do not weigh MOV. You cannot use the rule to justify the reason.

Just like in chess (where Elo ratings are most commonly used), all that matters is who wins.

One of our points of contention. Agree to disagree here. The fact that Michigan lost at Ohio State 39-42 means something entirely different than that Michigan lost at Ohio State by... 14-28, 0-42, etc. MOV reflects (to some extent) a measure of the relative parity of the two teams playing each other.


The predictor rankings (according to Sagarin) are better at guessing who would win in a matchup, but (and remember this next part), that isn't how teams should be ranked. We aren't asking "who is the best team" when we rank teams, we are asking "who has had the best season, according to the results of their games."

As far as I'm aware, there is no central question that polls are meant to answer. I have this discussion with friends every year -- what are polls supposed to rank? A measure of past accomplishments? A prediction of potential success? Something in between? This was incredibly pronounced for that brief point in time where Dennis Dixon was hobbled, but Oregon was still undefeated. I'm not sure that poll voters knew what to do about the Ducks. Were they supposed to look at the Oregon that had crushed everybody, or imagine the sputtering Oregon without Dixon?


Edit to add: Allowing nonsensical polls such as this is precisely why I don't participate in this poll. It's a complete waste of time.

Far from a waste of time. I would welcome entries that subverted the results of my ranking, because that would provide the opportunity to learn what criteria are involved in producing that kind of result. Aside from that, the law of averages is against whatever aberrational results my rankings churn out. I've noticed very few instances where the extra 10-15 points a team like A&M or Notre Dame was awarded, or where Arkansas or Kansas State was not, resulted in a change of position in the aggregated poll. Again, the more factors that are involved in producing a result, the more completely the result will reflect its inputs.

[MJ4H]

deleted a bunch of stuff that tries to justify poor rankings


Look, this all boils down to, as I can see by your last few sentences, the fact that you are trying to answer the wrong question. Of the two questions I posed, "what is the best team" and "what team has had the best season," only one of them makes sense for these rankings. If we attempted to answer the first question instead of the second question, the games themselves wouldn't actually matter. We would be allowed to think ridiculous things like "well even though Texas Tech beat Oklahoma, Oklahoma is still a better team, so I'm going to rank them as if they didn't lose that game. After all, they are better." Why did they even play?

This is just an example of the horrible results that can occur when you ask the wrong question. The ONLY thing that should matter is the results on the field. Otherwise, there is no point to having results on the field.

Furthermore, the only on-field result that matters is W or L. Otherwise, you fool with the integrity of the game. Teams start doing things like trying to win by as much as possible or trying to gain meaningless yards in games where the actual requirements of the game are to have more points than the other.

Understand: number of yards gained don't matter, number of yards allowed don't matter, number of points scored minus number of points allowed doesn't matter. What matters is points > or < points allowed. That is all as defined by the rules of the game, and that is all that should be assessed. Using other things fundamentally alters the game.

And I fully believe that your polls and others like it make the activity a total waste of time. I'm absolutely baffled that you continue to try to defend it.

edit to add:
There is value to things that do similar things to what you are trying to do, incidentally. Just not in ranking teams' season performance. It is noted that Sagarin predictor ratings do a pretty good job of predicting future results. That is certainly useful. BUT, it is also absolutely not what these rankings are supposed to be doing.

[JonInMiddleGA]

it is also absolutely not what these rankings are supposed to be doing.

I didn't know there was a set criteria for what rankings are "supposed" to do.

To some extent, "good losses" & "bad wins" weigh on most human pollsters. Otherwise, we'd just have lists in pure rank order based on wins, and Houston would be no worse than #3.

edit to add: Point being, while I think the poll under discussion goes overboard for my taste in how it weights some of the factors, it's unrealistic to pretend that aspects of what it's doing aren't present in a lot of polls.

[MJ4H]

I didn't know there was a set criteria for what rankings are "supposed" to do.

To some extent, "good losses" & "bad wins" weigh on most human pollsters. Otherwise, we'd just have lists in pure rank order based on wins, and Houston would be no worse than #3.

edit to add: Point being, while I think the poll under discussion goes overboard for my taste in how it weights some of the factors, it's unrealistic to pretend that aspects of what it's doing aren't present in a lot of polls.

Well, I think it is pretty clear what the polls are supposed to be doing, personally, if for no other reason than one of the two options leads to absurdity while the other follows the same logic as every other sport (other sports use criteria like win/loss, head-to-head, etc. to decide whose season was the best...they don't ignore that info and decide one team was better despite that info).

And I certainly realize that a lot of polls are rife with confusion about what they should be doing. It's one of my two major complaints about the whole sport. I'm quite certain a lot of human voters do the same thing.

[cartman]

The Top 25 from Solecismic's Football Frontier:

1. LSU
2. Alabama
3. Oklahoma
4. Boise State
5. Oregon
6. Oklahoma State
7. Stanford
8. Michigan
9. Georgia
10. Texas
11. Virginia Tech
12. South Carolina
13. USC
14. Arkansas
15. Houston
16. Southern Mississippi
17. Kansas State
18. TCU
19. Penn State
20. Clemson
21. Wisconsin
22. Nebraska
23. Utah
24. Michigan State
25. Missouri

[cartman]

bump...

you have until kickoff of the North Carolina-Va Tech game to get polls submitted

Here's my Top 25:

1. LSU
2. Oklahoma State
3. Alabama
4. Oregon
5. Arkansas
6. Virginia Tech
7. Oklahoma
8. Stanford
9. Boise State
10. Clemson
11. Houston
12. Nebraska
13. South Carolina
14. Michigan State
15. Wisconsin
16. Georgia
17. Kansas State
18. USC
19. Southern Mississippi
20. Penn State
21. Michigan
22. West Virginia
23. Florida State
24. Georgia Tech
25. Texas