I wondered if FCS polls were more accurate so I checked the Any Given Saturday (AGS) pre and final season polls for last year. Here is what I found-

14/25 teams ranked in the preseason were still around at season’s end, so a little better than the FBS poll.

4 top 10 FCS teams were missing from the final top 25, so slightly worse than the FBS poll.

At both levels, the accuracy of preseason polls is questionable.

IIRC, teams that start ranked finished ranked something like 70% or 75% of the time. It’s certainly better than if you picked teams at random out of a hat. (Which is literally what the guy on tiptop25 does every year in order to get his preseason rankings.)



However, the preseason rankings probably are a self-fulfilling prophecy to some extent. Part of why teams that start out ranked highly end ranked highly is just because they were ranked highly in the preseason.

(08-10-2022 12:02 PM)Poster Wrote:  IIRC, teams that start ranked finished ranked something like 70% or 75% of the time. It’s certainly better than if you picked teams at random out of a hat. (Which is literally what the guy on tiptop25 does every year in order to get his preseason rankings.)



However, the preseason rankings probably are a self-fulfilling prophecy to some extent. Part of why teams that start out ranked highly end ranked highly is just because they were ranked highly in the preseason.

The first post here shows it was 44% stayed ranked. Are you saying last year was an aberration or is this potentially just an overestimate on your part?

Checking 2019, it looks like it was 64% stayed ranked. 2018 was 56%. 2017 looks like 60%.

I skipped 2020 because it was obviously a crazy year and bad to judge off of.

But looks like it's probably closer to 60% staying ranked, without tracking how far off the rankings tend to be.

(08-10-2022 12:53 PM)e-parade Wrote:  

(08-10-2022 12:02 PM)Poster Wrote:  IIRC, teams that start ranked finished ranked something like 70% or 75% of the time. It’s certainly better than if you picked teams at random out of a hat. (Which is literally what the guy on tiptop25 does every year in order to get his preseason rankings.)



However, the preseason rankings probably are a self-fulfilling prophecy to some extent. Part of why teams that start out ranked highly end ranked highly is just because they were ranked highly in the preseason.

The first post here shows it was 44% stayed ranked. Are you saying last year was an aberration or is this potentially just an overestimate on your part?

Checking 2019, it looks like it was 64% stayed ranked. 2018 was 56%. 2017 looks like 60%.

I skipped 2020 because it was obviously a crazy year and bad to judge off of.

But looks like it's probably closer to 60% staying ranked, without tracking how far off the rankings tend to be.

Yeah, I didn’t read the OP.


I thought I had read before that about 70% or 75% of teams that start ranked finished ranked in a typical year. I might be remembering incorrectly.

(08-10-2022 12:56 PM)Poster Wrote:  

(08-10-2022 12:53 PM)e-parade Wrote:  

(08-10-2022 12:02 PM)Poster Wrote:  IIRC, teams that start ranked finished ranked something like 70% or 75% of the time. It’s certainly better than if you picked teams at random out of a hat. (Which is literally what the guy on tiptop25 does every year in order to get his preseason rankings.)



However, the preseason rankings probably are a self-fulfilling prophecy to some extent. Part of why teams that start out ranked highly end ranked highly is just because they were ranked highly in the preseason.

The first post here shows it was 44% stayed ranked. Are you saying last year was an aberration or is this potentially just an overestimate on your part?

Checking 2019, it looks like it was 64% stayed ranked. 2018 was 56%. 2017 looks like 60%.

I skipped 2020 because it was obviously a crazy year and bad to judge off of.

But looks like it's probably closer to 60% staying ranked, without tracking how far off the rankings tend to be.

I thought I had read before that about 70% or 75% of teams that start ranked finished ranked in a typical year. I might be remembering incorrectly.

If there is a trend making the preseason top 25 more inaccurate, it may be partly due to the fact that more and more non-P5 teams have been making their way into the final post-season top 25 in the past few years.

......................Number of
......................Non-P5 Teams
......................in the Final
......................AP Top 25
2014.....................3
2015.....................3
2016.....................3
2017.....................4
2018.....................6
2019.....................7
2020.....................8
2021.....................6

.

(08-09-2022 08:42 PM)CoastalJuan Wrote:  

(08-09-2022 07:45 PM)quo vadis Wrote:  

(08-09-2022 06:57 PM)jsw3ent Wrote:  The polls are rigged/biased against the little guy

I think they are biased in favor of the G5, at least between positions say 15-25. Compare the final polls with the final MC last year.

Voters seem to prefer good records vs soft schedules, so often vote a 10-2 G5 over an 8-4 P that played a much tougher schedule.

Considering the fact that the final coaches poll had zero 10-win G5s and one 8-4 P5, I challenge your assertion.

I should have been clearer - "Milwaukee" caught the drift of what I was trying to convey, meaning I didn't mean to literally label exactly 10 wins as what I was getting at. But that's on me, I can't expect you to be a mind-reader, so I apologize.

That said, if we are being specific, note that I did say "10-2" G5, not just "10 win". Looking at last year's standings, I don't see any G5 that finished with a 10-2 record. There were a couple teams, Fresno State and App State, that had 10 wins, but they had more losses, 3 and 4 respectively, so not precisely what I mentioned either. Again, that's because I didn't intend to zero in specifically on those records (and again, not your fault for thinking I did, as it is my job to be clear and I wasn't), the idea was to convey the general sense that a soft-schedule G5 with a good record - let's say "double digits" with two or fewer losses, is likely to get the nod over P5 that have played tougher schedules but have 4-5 losses. It's an approximation.

In any event, I think my assertion, independent of the specific records I noted as examples, the notion that voters prefer good records with soft schedules over worse records vs tough schedules, which works to the benefit of G5, is supported. If we look at the Massey Composite from last year, it had three G5 schools in its final top 25, compared to six in the AP (counting BYU as a "G") and five in the Coaches poll, and, in the case of the three that made the MC top 25, they were all ranked lower in the MC compared to their rankings in the human polls. For example, Houston was #17 in the final Coaches and AP poll, but #21 in the final MC rankings.

So to me, that is evidence that the pollsters tend to over-rate G5 with good records at the end of the year.

.

These data suggest that some of the inaccuracy/unreliability of the preseason top 25 may have been due to the fact that more non-P5 teams have earned spots in the final top 25 in recent years.

......................Number of................Number of
......................Non-P5 Teams..........Non-P5 Teams
......................in the Preseason.......in the Final
......................AP Top 25................AP Top 25..........Discrepancy:
2014.....................0............................3.......................-3 (100%)
2015.....................1............................3.......................-2 (66.7%)
2016.....................1............................3.......................-2 (66.7%)
2017.....................1............................4.......................-3 (75%)
2018.....................2............................6.......................-4 (66.7%)
2019.....................1............................7.......................-6 (85.7%)
2020.....................2............................8.......................-6 (75%)
2021.....................3............................6.......................-3 (50%)

The "hit rate" for the preseason top 25 appears to have been lower with respect to the non-P5 teams than it has been for the P5 teams.

.

I do agree with Quo (and some others) that pollsters tend to overrate (at least somewhat) G5 programs with strong records and questionable schedules — and perhaps even to the detriment of P5 programs with average records and demanding records.

And I am 100 percent fine with that.

(08-10-2022 02:38 PM)bill dazzle Wrote:  I do agree with Quo (and some others) that pollsters tend to overrate (at least somewhat) G5 programs with strong records and questionable schedules — and perhaps even to the detriment of P5 programs with average records and demanding records.

And I am 100 percent fine with that.

Agreed. It's mostly the late teens through 25 that would be impacted by this, and those don't do much in terms of the playoffs or elite bowls and such. It's a good reward for the non P5 teams that did what they could with the schedule they had, as opposed to a P5 team that likely underperformed their way to too many losses.

(08-10-2022 02:38 PM)bill dazzle Wrote:  I do agree with Quo (and some others) that pollsters tend to overrate (at least somewhat) G5 programs with strong records and questionable schedules — and perhaps even to the detriment of P5 programs with average records and demanding records.

And I am 100 percent fine with that.

About the bolded, I feel pretty much the same - it doesn't bother me when a 11-2 G5 that played a cupcake schedule and whose Big Bowl Win is over a 6-6 P5 team in the Birmingham Bowl or the Pizza Bowl gets ranked #22 while say an 8-5 P5 team that played a much harder one, and who the computers say was better, is left out.

But I do think that bias is there, and is worth noting.

(08-10-2022 02:59 PM)quo vadis Wrote:  

(08-10-2022 02:38 PM)bill dazzle Wrote:  I do agree with Quo (and some others) that pollsters tend to overrate (at least somewhat) G5 programs with strong records and questionable schedules — and perhaps even to the detriment of P5 programs with average records and demanding records.

And I am 100 percent fine with that.

About the bolded, I feel pretty much the same - it doesn't bother me when a 11-2 G5 that played a cupcake schedule and whose Big Bowl Win is over a 6-6 P5 team in the Birmingham Bowl or the Pizza Bowl gets ranked #22 while say an 8-5 P5 team that played a much harder one, and who the computers say was better, is left out.

But I do think that bias is there, and is worth noting.

Such a thinly veiled shot a UH and it clearly does bother you lmao

(08-10-2022 03:08 PM)WhoseHouse? Wrote:  

(08-10-2022 02:59 PM)quo vadis Wrote:  

(08-10-2022 02:38 PM)bill dazzle Wrote:  I do agree with Quo (and some others) that pollsters tend to overrate (at least somewhat) G5 programs with strong records and questionable schedules — and perhaps even to the detriment of P5 programs with average records and demanding records.

And I am 100 percent fine with that.

About the bolded, I feel pretty much the same - it doesn't bother me when a 11-2 G5 that played a cupcake schedule and whose Big Bowl Win is over a 6-6 P5 team in the Birmingham Bowl or the Pizza Bowl gets ranked #22 while say an 8-5 P5 team that played a much harder one, and who the computers say was better, is left out.

But I do think that bias is there, and is worth noting.

Such a thinly veiled shot a UH and it clearly does bother you lmao

Eh, it's fun taking shots at Houston, LOL. But yes, their 2021 bowl came to mind, as an example. Though in a seasonal sense, not a good one, as the MC did have Houston ranked #21.

(08-10-2022 03:15 PM)quo vadis Wrote:  

(08-10-2022 03:08 PM)WhoseHouse? Wrote:  

(08-10-2022 02:59 PM)quo vadis Wrote:  

(08-10-2022 02:38 PM)bill dazzle Wrote:  I do agree with Quo (and some others) that pollsters tend to overrate (at least somewhat) G5 programs with strong records and questionable schedules — and perhaps even to the detriment of P5 programs with average records and demanding records.

And I am 100 percent fine with that.

About the bolded, I feel pretty much the same - it doesn't bother me when a 11-2 G5 that played a cupcake schedule and whose Big Bowl Win is over a 6-6 P5 team in the Birmingham Bowl or the Pizza Bowl gets ranked #22 while say an 8-5 P5 team that played a much harder one, and who the computers say was better, is left out.

But I do think that bias is there, and is worth noting.

Such a thinly veiled shot a UH and it clearly does bother you lmao

Eh, it's fun taking shots at Houston, LOL. But yes, their 2021 season came to mind, as an example.

Haha I respect you for owning it and yes our schedule last year was especially bad

(08-10-2022 03:18 PM)WhoseHouse? Wrote:  

(08-10-2022 03:15 PM)quo vadis Wrote:  

(08-10-2022 03:08 PM)WhoseHouse? Wrote:  

(08-10-2022 02:59 PM)quo vadis Wrote:  

(08-10-2022 02:38 PM)bill dazzle Wrote:  I do agree with Quo (and some others) that pollsters tend to overrate (at least somewhat) G5 programs with strong records and questionable schedules — and perhaps even to the detriment of P5 programs with average records and demanding records.

And I am 100 percent fine with that.

About the bolded, I feel pretty much the same - it doesn't bother me when a 11-2 G5 that played a cupcake schedule and whose Big Bowl Win is over a 6-6 P5 team in the Birmingham Bowl or the Pizza Bowl gets ranked #22 while say an 8-5 P5 team that played a much harder one, and who the computers say was better, is left out.

But I do think that bias is there, and is worth noting.

Such a thinly veiled shot a UH and it clearly does bother you lmao

Eh, it's fun taking shots at Houston, LOL. But yes, their 2021 season came to mind, as an example.

Haha I respect you for owning it and yes our schedule last year was especially bad

Well, we, USF, definitely played our part in making it bad, LOL. Hopefully this year we will be better.

(08-10-2022 02:32 PM)Milwaukee Wrote:  .

These data suggest that some of the inaccuracy/unreliability of the preseason top 25 may have been due to the fact that more non-P5 teams have earned spots in the final top 25 in recent years.

......................Number of................Number of
......................Non-P5 Teams..........Non-P5 Teams
......................in the Preseason.......in the Final
......................AP Top 25................AP Top 25..........Discrepancy:
2014.....................0............................3.......................-3 (100%)
2015.....................1............................3.......................-2 (66.7%)
2016.....................1............................3.......................-2 (66.7%)
2017.....................1............................4.......................-3 (75%)
2018.....................2............................6.......................-4 (66.7%)
2019.....................1............................7.......................-6 (85.7%)
2020.....................2............................8.......................-6 (75%)
2021.....................3............................6.......................-3 (50%)

The "hit rate" for the preseason top 25 appears to have been lower with respect to the non-P5 teams than it has been for the P5 teams.

.

If you go further back in time to about the 2006-11 period, you’d find more non power conference teams then you did from 2014-17.


Of course, the difference is that the MWC and WAC teams who finished ranked in that period actually were legitimate top 25 teams. Nowadays, it’s usually just a high ranking because the team didn’t have very many losses.


I also suspect that there wasn’t such a discrepancy between non-power teams that were ranked in the preseason and finished ranked in the final poll back then. Teams like TCU, Utah and Boise State were generally in the preseason top 25 because people actually thought they were among the best 25 teams in the country. I think the preseason poll usually does reflect who the coaches and writers honestly think are the best 25 teams in the country. But as the year goes on, the poll comes to largely just reflect who has the least losses.

(08-10-2022 01:58 PM)quo vadis Wrote:  

(08-09-2022 08:42 PM)CoastalJuan Wrote:  

(08-09-2022 07:45 PM)quo vadis Wrote:  

(08-09-2022 06:57 PM)jsw3ent Wrote:  The polls are rigged/biased against the little guy

I think they are biased in favor of the G5, at least between positions say 15-25. Compare the final polls with the final MC last year.

Voters seem to prefer good records vs soft schedules, so often vote a 10-2 G5 over an 8-4 P that played a much tougher schedule.

Considering the fact that the final coaches poll had zero 10-win G5s and one 8-4 P5, I challenge your assertion.

I should have been clearer - "Milwaukee" caught the drift of what I was trying to convey, meaning I didn't mean to literally label exactly 10 wins as what I was getting at. But that's on me, I can't expect you to be a mind-reader, so I apologize.

That said, if we are being specific, note that I did say "10-2" G5, not just "10 win". Looking at last year's standings, I don't see any G5 that finished with a 10-2 record. There were a couple teams, Fresno State and App State, that had 10 wins, but they had more losses, 3 and 4 respectively, so not precisely what I mentioned either. Again, that's because I didn't intend to zero in specifically on those records (and again, not your fault for thinking I did, as it is my job to be clear and I wasn't), the idea was to convey the general sense that a soft-schedule G5 with a good record - let's say "double digits" with two or fewer losses, is likely to get the nod over P5 that have played tougher schedules but have 4-5 losses. It's an approximation.

In any event, I think my assertion, independent of the specific records I noted as examples, the notion that voters prefer good records with soft schedules over worse records vs tough schedules, which works to the benefit of G5, is supported. If we look at the Massey Composite from last year, it had three G5 schools in its final top 25, compared to six in the AP (counting BYU as a "G") and five in the Coaches poll, and, in the case of the three that made the MC top 25, they were all ranked lower in the MC compared to their rankings in the human polls. For example, Houston was #17 in the final Coaches and AP poll, but #21 in the final MC rankings.

So to me, that is evidence that the pollsters tend to over-rate G5 with good records at the end of the year.

There is a very good and practical reason for that. For most of the history of football polls (starting in 1936) The polls didn't matter very much for most schools. For a long time, they were finished at the end of the regular season and didn't even count the bowl games, which were exhibitions. They started out only ranking the Top Ten teams. Until the start of the BCS they had no impact on who played whom in the postseason.

Eventually, they grew to a Top 25 to help out more coaches who got bonuses for finishing ranked or getting "also receiving votes" status. But still, only a handful of teams were actually affected by their ranking. It wasn't until the onset of the CFP in which 12 teams' bowl placement was dictated by ranking that we started to see more G5 schools getting ranked. Why? Because there was an AQ given to the G5 champ ranked the highest. Since there is no separate ranking of G5 teams (maybe there should be) voters took it on themselves to speculate on who should get that coveted access slot (perhaps influencing the selection committee who announced their rankings after the AP and Coaches polls).

As far as preseason polls, I'd be inclined to bet that if you made the final poll from one year the preseason poll for the next, you would get roughly the same number of schools that were still ranked at the end of the year. Clearly, the current system isn't based on any valid analysis of team rosters, returning production, coaching changes or schedules. Voters are by and large pulling their picks out of a bodily orifice. Anyone here could do as well.

(08-10-2022 02:38 PM)bill dazzle Wrote:  I do agree with Quo (and some others) that pollsters tend to overrate (at least somewhat) G5 programs with strong records and questionable schedules — and perhaps even to the detriment of P5 programs with average records and demanding records.

And I am 100 percent fine with that.

It may be the case that there is an implicit/psychological "threshold" of 10 wins that helps put a team into the top 25, much like there used to be (and may still be, to some extent) an implicit threshold of 20 wins that once helped teams earn NCAA bids. Notably, every P5 team that won 10+ games ended up in the final top 25 last season.

To the extent that it's slightly easier for some non-P5 teams to win 10 games, that could give them a boost with some poll voters. However, 5 of the 11 non-P5 schools that won 10+ games last season didn't make it into the final AP top 25, proving that simply winning 10+ games wasn't enough to get a non-P5 team into the final top 25.

. . . . . . . . . . . . .

Let's look at the marginal non-P5 top 25 teams - - those that were ranked #24th and #25th:

Q: Why did #24 Utah State end up in the final AP Top 25?

A: Utah State beat (P5) Washington State, (P5) Oregon State, (10-3) Air Force, and (12-2) SDSU in the MWC championship game, and finished with a 2-0 record vs. P5 teams.

Q: Why did #25 SDSU end up in the final AP Top 25?

A: SDSU beat (#12; 10-4) Utah, (P5) Arizona, (10-3) Air Force, and (12-2) UTSA, and finished with a 2-0 record vs. P5 teams.

. . . . . . . . . . . . .


Now, consider #23 (10-4) Iowa's record:

Sat 01-01 vs. (10-3) Kentucky L
Sat 12-04 vs (12-2) Michigan L
Fri 11-26 at Nebraska (3-9) W
Sat 11-20 Illinois (5-7) W
Sat 11-13 Minnesota (9-4) W
Sat 11-06 at Northwestern (3-9) W
Sat 10-30 at Wisconsin (9-4) L
Sat 10-16 Purdue (9-4) L
Sat 10-09 Penn St (7-6) W
Fri 10-01 at Maryland (7-6) W
Sat 09-25 Colorado St (3-9) W
Sat 09-18 Kent (7-7) W
Sat 09-11 at Iowa St (7-6) W
Sat 09-04 Indiana (2-10) W

In summary, Iowa's major accomplishments in 2021 were that they beat one 9-4 team and two 7-6 teams, yet Iowa ended up being ranked ahead of Utah State and SDSU in the final AP top 25.

This example illustrates the point that, while it's possible there may be some "double digit win" bias operating in favor of certain non-P5 teams, there may also be some "pro-P5" bias operating in favor of certain P5 teams, such as Iowa in 2021.

.

(08-10-2022 01:58 PM)quo vadis Wrote:  

(08-09-2022 08:42 PM)CoastalJuan Wrote:  

(08-09-2022 07:45 PM)quo vadis Wrote:  

(08-09-2022 06:57 PM)jsw3ent Wrote:  The polls are rigged/biased against the little guy

I think they are biased in favor of the G5, at least between positions say 15-25. Compare the final polls with the final MC last year.

Voters seem to prefer good records vs soft schedules, so often vote a 10-2 G5 over an 8-4 P that played a much tougher schedule.

Considering the fact that the final coaches poll had zero 10-win G5s and one 8-4 P5, I challenge your assertion.

I should have been clearer - "Milwaukee" caught the drift of what I was trying to convey, meaning I didn't mean to literally label exactly 10 wins as what I was getting at. But that's on me, I can't expect you to be a mind-reader, so I apologize.

That said, if we are being specific, note that I did say "10-2" G5, not just "10 win". Looking at last year's standings, I don't see any G5 that finished with a 10-2 record. There were a couple teams, Fresno State and App State, that had 10 wins, but they had more losses, 3 and 4 respectively, so not precisely what I mentioned either. Again, that's because I didn't intend to zero in specifically on those records (and again, not your fault for thinking I did, as it is my job to be clear and I wasn't), the idea was to convey the general sense that a soft-schedule G5 with a good record - let's say "double digits" with two or fewer losses, is likely to get the nod over P5 that have played tougher schedules but have 4-5 losses. It's an approximation.

In any event, I think my assertion, independent of the specific records I noted as examples, the notion that voters prefer good records with soft schedules over worse records vs tough schedules, which works to the benefit of G5, is supported. If we look at the Massey Composite from last year, it had three G5 schools in its final top 25, compared to six in the AP (counting BYU as a "G") and five in the Coaches poll, and, in the case of the three that made the MC top 25, they were all ranked lower in the MC compared to their rankings in the human polls. For example, Houston was #17 in the final Coaches and AP poll, but #21 in the final MC rankings.

So to me, that is evidence that the pollsters tend to over-rate G5 with good records at the end of the year.

I get all that. I was just using an example when I mentioned Texas A&M.

My point was more that the handicap that P5 teams are getting seems to be more than 2 wins.

Examples from AP Poll: 100% of 10-win P5s got in, vs. 60% of G5s with 12-win records. The math holds down the line for any 2-game difference. Stands to reason that the P5 gimme is more than 2. It might be less than 3, but it's more than 2.

G5 12 (60%) vs. P5 10 (100%)
Houston - In
UTSA - Out
SDSU - In
App State - Out
BYU - In
Clemson - In
Iowa - In
Oregon - In
Utah - In
Tucky - In
Ole Miss - In