(10-15-2025 07:38 PM)ken d Wrote:  

(10-15-2025 05:17 PM)YNot Wrote:  Some valid points. Totally agree that LSU is overrated and rankings inertia is a real thing. The recent admissions of pollsters that they didn't observe a particular result or didn't feel that their vote really matters discredit the polls and the rankings systems.

The USC and Illinois examples are poorly selected. Illinois didn't just lose to one of the best teams, they lost by 53 points! That matters. And, what is USC's big win? Michigan? Why was Michigan ranked? Because of a solid 11-point road loss to Oklahoma? (who didn't look so hot on Saturday) Or, was it because of Michigan's brand and conference affiliation?

And it was the Big Ten's own Penn State that made a mockery of the polls by sitting in the TOP-TWO for a month, without a single P4 win...because of.....brand and conference affiliation...only to nosedive into mediocrity. Seems like that should have been a primary exhibit to support his thesis that the polls are inherently biased, but he conveniently left them out because of....his own biases.

We all do it. And with so few cross-conference games, it's extremely subjective and often based on the eye test. I support a reasonable increase in the number of autobids. Settle things with actual games, not by beauty pageant.

I have never made a secret of the fact that I dislike polls. All polls.

Polls and ranking algorithms have two things in common. They are all flawed and they are all biased. When it comes to college football we have never really had a playoff that had a field that everybody agreed was the "right" one. That isn't to say we have never had a "champion" that was overwhelmingly believed to be the "best" team that year after the playoff ended.

We all know that, in other college sports, the winner of the national championship tournament is not always the best team. But their championship is not disputed, because the nature of those sports is such that all teams capable of winning the tournament have a path to enter the field, even though some teams in the field who have no chance of winning are also included.

Now that we have a larger CFP field we are getting close to a point where there aren't any teams capable of winning it that are excluded. But there are still likely some teams who would be capable of upsetting a team that could be the champion who don't get a bid. Because of the money involved in just getting invited, flawed polls or other subjective processes for selecting teams, there needs to be a way, IMO, for selecting teams based on performance on the field, not in polls.

Even though it is subjective in a way to award more autobids to some conferences than others, I still prefer that to polls or committees. Very few people involved in or fans of college football would dispute that the SEC and B1G have more resources and more good teams than any other conferences, and that the other two P conferences have more resources and more good teams than G conferences do. That should be reflected in how many autobids each conference gets.

I believe that the best way to decide who gets to play in the CFP is by having CCGs and WCGs among teams based on where they finish in their conference standings, regardless of where those teams are ranked in human polls or computer algorithms. I hope we are moving in that direction.

So you don't like the way the NCAA basketball tourney is selected? Over half the teams are decided by a committee, just like the CFP.

(10-15-2025 07:38 PM)ken d Wrote:  

(10-15-2025 05:17 PM)YNot Wrote:  Some valid points. Totally agree that LSU is overrated and rankings inertia is a real thing. The recent admissions of pollsters that they didn't observe a particular result or didn't feel that their vote really matters discredit the polls and the rankings systems.

The USC and Illinois examples are poorly selected. Illinois didn't just lose to one of the best teams, they lost by 53 points! That matters. And, what is USC's big win? Michigan? Why was Michigan ranked? Because of a solid 11-point road loss to Oklahoma? (who didn't look so hot on Saturday) Or, was it because of Michigan's brand and conference affiliation?

And it was the Big Ten's own Penn State that made a mockery of the polls by sitting in the TOP-TWO for a month, without a single P4 win...because of.....brand and conference affiliation...only to nosedive into mediocrity. Seems like that should have been a primary exhibit to support his thesis that the polls are inherently biased, but he conveniently left them out because of....his own biases.

We all do it. And with so few cross-conference games, it's extremely subjective and often based on the eye test. I support a reasonable increase in the number of autobids. Settle things with actual games, not by beauty pageant.

I have never made a secret of the fact that I dislike polls. All polls.

Polls and ranking algorithms have two things in common. They are all flawed and they are all biased. When it comes to college football we have never really had a playoff that had a field that everybody agreed was the "right" one. That isn't to say we have never had a "champion" that was overwhelmingly believed to be the "best" team that year after the playoff ended.

We all know that, in other college sports, the winner of the national championship tournament is not always the best team. But their championship is not disputed, because the nature of those sports is such that all teams capable of winning the tournament have a path to enter the field, even though some teams in the field who have no chance of winning are also included.

Now that we have a larger CFP field we are getting close to a point where there aren't any teams capable of winning it that are excluded. But there are still likely some teams who would be capable of upsetting a team that could be the champion who don't get a bid. Because of the money involved in just getting invited, flawed polls or other subjective processes for selecting teams, there needs to be a way, IMO, for selecting teams based on performance on the field, not in polls.

Even though it is subjective in a way to award more autobids to some conferences than others, I still prefer that to polls or committees. Very few people involved in or fans of college football would dispute that the SEC and B1G have more resources and more good teams than any other conferences, and that the other two P conferences have more resources and more good teams than G conferences do. That should be reflected in how many autobids each conference gets.

I believe that the best way to decide who gets to play in the CFP is by having CCGs and WCGs among teams based on where they finish in their conference standings, regardless of where those teams are ranked in human polls or computer algorithms. I hope we are moving in that direction.

The CFP isn’t really about determining a champion, not really. It’s about making craploads of money for the power schools. It just so happens that lots of people like to watch playoff games, and every CFP thus far has led to a champ that both the AP and coaches polls agreed on. Hardly a guarantee though, is it? Would a 3 loss CFP champ be #1 in the polls ahead of a 1 loss team that won the B1G, was ranked #1 all year, and lost in a nail biter in the Title game? Say OSU is that 1 loss team, and Texas is the 3 loss team, making it even more confusing bc the 2 teams split their series for the year? Hardly, OSU would easily be #1 in the AP and Coaches in that scenario, and we’d have another split Title. A huge side benefit of the CFP is all the controversy created by letting some schools in and keeping some schools out. Gotta keep the 24/7 news cycle fed or else the talking heads will find other things to yell and argue about.

Limiting the CFP to 4 schools was a good way to determine a champion. Allowing in 12++ is a good way to make everybody a lot of money.

(10-15-2025 07:38 PM)ken d Wrote:  I believe that the best way to decide who gets to play in the CFP is by having CCGs and WCGs among teams based on where they finish in their conference standings, regardless of where those teams are ranked in human polls or computer algorithms. I hope we are moving in that direction.

Conference standings are also algorithm generated, only they ignore a quarter of the games played. They are “fair” because they are transparent. Use a transparent computer algorithm to rank all teams based on all games and you have a winning combination.

This was the original goal of the BCS, but we hadn’t reached nerd-critical-mass in 1998 and implemented it in a poor way.

(10-16-2025 05:30 AM)Crayton Wrote:  

(10-15-2025 07:38 PM)ken d Wrote:  I believe that the best way to decide who gets to play in the CFP is by having CCGs and WCGs among teams based on where they finish in their conference standings, regardless of where those teams are ranked in human polls or computer algorithms. I hope we are moving in that direction.

Conference standings are also algorithm generated, only they ignore a quarter of the games played. They are “fair” because they are transparent. Use a transparent computer algorithm to rank all teams based on all games and you have a winning combination.

This was the original goal of the BCS, but we hadn’t reached nerd-critical-mass in 1998 and implemented it in a poor way.

And given the large number of teams in conferences and the resulting uneven scheduling, conference standings aren't dramatically more "fair." Especially as you get below the champion who actually won a ccg.

(10-16-2025 09:04 AM)bullet Wrote:  

(10-16-2025 05:30 AM)Crayton Wrote:  

(10-15-2025 07:38 PM)ken d Wrote:  I believe that the best way to decide who gets to play in the CFP is by having CCGs and WCGs among teams based on where they finish in their conference standings, regardless of where those teams are ranked in human polls or computer algorithms. I hope we are moving in that direction.

Conference standings are also algorithm generated, only they ignore a quarter of the games played. They are “fair” because they are transparent. Use a transparent computer algorithm to rank all teams based on all games and you have a winning combination.

This was the original goal of the BCS, but we hadn’t reached nerd-critical-mass in 1998 and implemented it in a poor way.

And given the large number of teams in conferences and the resulting uneven scheduling, conference standings aren't dramatically more "fair." Especially as you get below the champion who actually won a ccg.

I don't believe I said using conference standings was fair. With the very large conferences we have today, no one can come close to a round robin schedule, which means some teams will have more favorable schedules than others. And I don't believe they are algorithm generated. The teams are ranked by their conference W-L record and various tie breakers that are mostly also based on records against other teams in the conference. The larger point is that they aren't determined by a committee outside the conference. If the conference were to get multiple autobids, they should be able to decide which of their members gets those bids.

If the polls devolve into 20 P2 schools and five for everyone else, they will become irrelevant.

(10-16-2025 09:23 AM)DFW HOYA Wrote:  If the polls devolve into 20 P2 schools and five for everyone else, they will become irrelevant.

Well if the Big Ten takes Miami and UVA, it would be 17 for the P2 based on today, 10 SEC and 7 Big Ten.

(10-16-2025 09:04 AM)bullet Wrote:  

(10-16-2025 05:30 AM)Crayton Wrote:  

(10-15-2025 07:38 PM)ken d Wrote:  I believe that the best way to decide who gets to play in the CFP is by having CCGs and WCGs among teams based on where they finish in their conference standings, regardless of where those teams are ranked in human polls or computer algorithms. I hope we are moving in that direction.

Conference standings are also algorithm generated, only they ignore a quarter of the games played. They are “fair” because they are transparent. Use a transparent computer algorithm to rank all teams based on all games and you have a winning combination.

This was the original goal of the BCS, but we hadn’t reached nerd-critical-mass in 1998 and implemented it in a poor way.

And given the large number of teams in conferences and the resulting uneven scheduling, conference standings aren't dramatically more "fair." Especially as you get below the champion who actually won a ccg.

For sure. The uneven scheduling in these large conferences will continue to result in rather arbitrary processes to select the CCG participants.

I would prefer to see the P4 conferences remove most of the silly payday home games and expand the CONFERENCE championship process and merge that into the CFP expansion somehow.

(10-16-2025 10:17 AM)bullet Wrote:  

(10-16-2025 09:23 AM)DFW HOYA Wrote:  If the polls devolve into 20 P2 schools and five for everyone else, they will become irrelevant.

Well if the Big Ten takes Miami and UVA, it would be 17 for the P2 based on today, 10 SEC and 7 Big Ten.

If the Big Ten takes Miami and UVA, UVA is probably not ranked and there's a decent chance that Miami would have knocked out one of the fringe B1G teams.

It's already a crowded field and someone else will step in to fill the void, like Virginia and Georgia Tech have done with Clemson's and Florida State's drop.

The American used to have Memphis, Houston, Cincinnati, and UCF at the top and vying for CFP rankings; USF and Tulane were bottom-feeders. Today, it's Memphis, *USF*, *Tulane*, and Navy.

(10-15-2025 08:13 PM)otown Wrote:  

(10-15-2025 04:41 PM)quo vadis Wrote:  

(10-15-2025 04:30 PM)otown Wrote:  

(10-15-2025 03:49 PM)bullet Wrote:  

(10-15-2025 02:32 PM)otown Wrote:  I know, common knowledge. However, even though I do feel SEC is the strongest conference, this piece does make me second guess some things. Now, before anyone says that these polls mean nothing and it is the CFP poll that really matters, I am quite confident these polls do influence the committee somewhat, especially their first ranking.
https://www.outkick.com/sports/college-f...disconnect

These systems he is using for his conclusion are pretty much a joke.

Regardless, the LSU example does hold merit.

FWIW, I think LSU is not likely to end the regular season as a ranked team. They are IMO a weakly-coached paper tiger - and I blame coaching because all the talent indices I am aware of say they have top 10 talent - and are likely to be out of the rankings three weeks from now. They play Vandy, Alabama and Texas AM and IMO will be lucky to win one of those. Brian Kelly could soon be sitting on the hottest seat in the country and LSU might be joining Penn State and Florida as elite programs searching for a new coach. Wouldn't surprise me.

That said, it's hard for me to agree that they are overrated right now. Both polls have them at #10 and ... the Massey Composite has them at #11 and Sagarin has them at #10. ESPN's "power index" has them at #11.

So IMO, there seems to be a very close congruence between the voters and the computers, no evidence that the voters are succumbing to a false narrative about them now.

Thats the point. They failed the current eye test, and you admit it too. Unfortunately when a poll has baked in bias, it picks and chooses when to enforce the eye test. Realistically, current resumes should hold more merit. Who care if they beat a ranked team early in the season if that team turned out to be hot garbage. Doesn't sagarin and massey still contain data from last year? Wouldn't that nullify them having any relevance at this point?

My understanding is that among the computers that use data from last year as filler, that typically is washed out by the sixth or seventh game of the season. So we're right about at the point where that data should be out of the system.

(10-16-2025 10:47 AM)quo vadis Wrote:  

(10-15-2025 08:13 PM)otown Wrote:  

(10-15-2025 04:41 PM)quo vadis Wrote:  

(10-15-2025 04:30 PM)otown Wrote:  

(10-15-2025 03:49 PM)bullet Wrote:  These systems he is using for his conclusion are pretty much a joke.

Regardless, the LSU example does hold merit.

FWIW, I think LSU is not likely to end the regular season as a ranked team. They are IMO a weakly-coached paper tiger - and I blame coaching because all the talent indices I am aware of say they have top 10 talent - and are likely to be out of the rankings three weeks from now. They play Vandy, Alabama and Texas AM and IMO will be lucky to win one of those. Brian Kelly could soon be sitting on the hottest seat in the country and LSU might be joining Penn State and Florida as elite programs searching for a new coach. Wouldn't surprise me.

That said, it's hard for me to agree that they are overrated right now. Both polls have them at #10 and ... the Massey Composite has them at #11 and Sagarin has them at #10. ESPN's "power index" has them at #11.

So IMO, there seems to be a very close congruence between the voters and the computers, no evidence that the voters are succumbing to a false narrative about them now.

Thats the point. They failed the current eye test, and you admit it too. Unfortunately when a poll has baked in bias, it picks and chooses when to enforce the eye test. Realistically, current resumes should hold more merit. Who care if they beat a ranked team early in the season if that team turned out to be hot garbage. Doesn't sagarin and massey still contain data from last year? Wouldn't that nullify them having any relevance at this point?

My understanding is that among the computers that use data from last year as filler, that typically is washed out by the sixth or seventh game of the season. So we're right about at the point where that data should be out of the system.

To be fair, LSU ranks highly in all computer rankings.

LSU is #13 in the Colley Matrix, which is a pure computer ranking based on who you played and who you beat. There's no preseason information in it, and it doesn't use margin of victory or even game location.

It says LSU beat #50, 68, 70, 72, and 142, and lost to #2.
In comparison, #12 Georgia beat #26, 57, 78, 98, and 116, and lost to #9.
#14 USF beat #29, 43, 72, 116, and 137, and lost to #5.

So it looks like LSU's advantage is that they only played one true cakewalk (which was FCS SE Louisiana)

(10-16-2025 12:16 PM)Captain Bearcat Wrote:  

(10-16-2025 10:47 AM)quo vadis Wrote:  

(10-15-2025 08:13 PM)otown Wrote:  

(10-15-2025 04:41 PM)quo vadis Wrote:  

(10-15-2025 04:30 PM)otown Wrote:  Regardless, the LSU example does hold merit.

FWIW, I think LSU is not likely to end the regular season as a ranked team. They are IMO a weakly-coached paper tiger - and I blame coaching because all the talent indices I am aware of say they have top 10 talent - and are likely to be out of the rankings three weeks from now. They play Vandy, Alabama and Texas AM and IMO will be lucky to win one of those. Brian Kelly could soon be sitting on the hottest seat in the country and LSU might be joining Penn State and Florida as elite programs searching for a new coach. Wouldn't surprise me.

That said, it's hard for me to agree that they are overrated right now. Both polls have them at #10 and ... the Massey Composite has them at #11 and Sagarin has them at #10. ESPN's "power index" has them at #11.

So IMO, there seems to be a very close congruence between the voters and the computers, no evidence that the voters are succumbing to a false narrative about them now.

Thats the point. They failed the current eye test, and you admit it too. Unfortunately when a poll has baked in bias, it picks and chooses when to enforce the eye test. Realistically, current resumes should hold more merit. Who care if they beat a ranked team early in the season if that team turned out to be hot garbage. Doesn't sagarin and massey still contain data from last year? Wouldn't that nullify them having any relevance at this point?

My understanding is that among the computers that use data from last year as filler, that typically is washed out by the sixth or seventh game of the season. So we're right about at the point where that data should be out of the system.

To be fair, LSU ranks highly in all computer rankings.

LSU is #13 in the Colley Matrix, which is a pure computer ranking based on who you played and who you beat. There's no preseason information in it, and it doesn't use margin of victory or even game location.

It says LSU beat #50, 68, 70, 72, and 142, and lost to #2.
In comparison, #12 Georgia beat #26, 57, 78, 98, and 116, and lost to #9.
#14 USF beat #29, 43, 72, 116, and 137, and lost to #5.

So it looks like LSU's advantage is that they only played one true cakewalk (which was FCS SE Louisiana)

I agree, the computers seem to be in accord in thinking LSU is a top 15 type team right now.

As an aside, I have no use for Colley-Matrix other than that it has the easiest OOC comparison matrix to read. Basically, in that computer, won-loss record, regardless of SOS, seems to be far and away the key determinant. They have Memphis at #11 right now, based on god knows what, for example, other than 6-0.

(10-16-2025 09:22 AM)ken d Wrote:  

(10-16-2025 05:30 AM)Crayton Wrote:  

(10-15-2025 07:38 PM)ken d Wrote:  Polls and ranking algorithms have two things in common. They are all flawed and they are all biased. […] Because of the money involved in just getting invited, flawed polls or other subjective processes for selecting teams, there needs to be a way, IMO, for selecting teams based on performance on the field, not in polls. Even though it is subjective in a way to award more autobids to some conferences than others, I still prefer that to polls or committees.

I believe that the best way to decide who gets to play in the CFP is by having CCGs and WCGs among teams based on where they finish in their conference standings, regardless of where those teams are ranked in human polls or computer algorithms. I hope we are moving in that direction.

Conference standings are also algorithm generated, only they ignore a quarter of the games played. They are “fair” because they are transparent. Use a transparent computer algorithm to rank all teams based on all games and you have a winning combination.

This was the original goal of the BCS, but we hadn’t reached nerd-critical-mass in 1998 and implemented it in a poor way.

I don't believe I said using conference standings was fair. With the very large conferences we have today, no one can come close to a round robin schedule, which means some teams will have more favorable schedules than others. And I don't believe they are algorithm generated. The teams are ranked by their conference W-L record and various tie breakers that are mostly also based on records against other teams in the conference. The larger point is that they aren't determined by a committee outside the conference. If the conference were to get multiple autobids, they should be able to decide which of their members gets those bids.

Ah, you implied “objective” which is probably the better word. I think that goes hand-in-hand with my use of “transparency”. And you are right, in either case, a human committee is deficient.

Tie-breakers are indeed algorithms. Each conference injects its unique blend of bias when building them. And I think it is probably an acceptable mechanism to allow conferences to independently determine how to allocate their bids to teams.

I would just push for the method of first allocating those bids to the conferences to likewise be built on an objective and transparent algorithm (rather than a roomfull of suits).

(10-16-2025 12:34 PM)quo vadis Wrote:  

(10-16-2025 12:16 PM)Captain Bearcat Wrote:  

(10-16-2025 10:47 AM)quo vadis Wrote:  

(10-15-2025 08:13 PM)otown Wrote:  

(10-15-2025 04:41 PM)quo vadis Wrote:  FWIW, I think LSU is not likely to end the regular season as a ranked team. They are IMO a weakly-coached paper tiger - and I blame coaching because all the talent indices I am aware of say they have top 10 talent - and are likely to be out of the rankings three weeks from now. They play Vandy, Alabama and Texas AM and IMO will be lucky to win one of those. Brian Kelly could soon be sitting on the hottest seat in the country and LSU might be joining Penn State and Florida as elite programs searching for a new coach. Wouldn't surprise me.

That said, it's hard for me to agree that they are overrated right now. Both polls have them at #10 and ... the Massey Composite has them at #11 and Sagarin has them at #10. ESPN's "power index" has them at #11.

So IMO, there seems to be a very close congruence between the voters and the computers, no evidence that the voters are succumbing to a false narrative about them now.

Thats the point. They failed the current eye test, and you admit it too. Unfortunately when a poll has baked in bias, it picks and chooses when to enforce the eye test. Realistically, current resumes should hold more merit. Who care if they beat a ranked team early in the season if that team turned out to be hot garbage. Doesn't sagarin and massey still contain data from last year? Wouldn't that nullify them having any relevance at this point?

My understanding is that among the computers that use data from last year as filler, that typically is washed out by the sixth or seventh game of the season. So we're right about at the point where that data should be out of the system.

To be fair, LSU ranks highly in all computer rankings.

LSU is #13 in the Colley Matrix, which is a pure computer ranking based on who you played and who you beat. There's no preseason information in it, and it doesn't use margin of victory or even game location.

It says LSU beat #50, 68, 70, 72, and 142, and lost to #2.
In comparison, #12 Georgia beat #26, 57, 78, 98, and 116, and lost to #9.
#14 USF beat #29, 43, 72, 116, and 137, and lost to #5.

So it looks like LSU's advantage is that they only played one true cakewalk (which was FCS SE Louisiana)

I agree, the computers seem to be in accord in thinking LSU is a top 15 type team right now.

As an aside, I have no use for Colley-Matrix other than that it has the easiest OOC comparison matrix to read. Basically, in that computer, won-loss record, regardless of SOS, seems to be far and away the key determinant. They have Memphis at #11 right now, based on god knows what, for example, other than 6-0.

Yeah. Colley was the worst of the BCS computer rankings.

This website tracks accuracy amongst the computers.

https://www.thepredictiontracker.com/ncaaresults.php

To summarize, the oddsmakers should be in charge of the polls as they have the most accurate software.