Home

Fixing the Final 2019 AP College Football Poll

1) Louisiana State 15-0
2) Clemson 14-1
3) Ohio State 13-1
4) Georgia 12-2
5) Oregon 12-2
6) Florida 11-2
7) Oklahoma 12-2
8) Alabama 11-2
9) Penn State 11-2
10) Minnesota 11-2
11) Wisconsin 10-4
12) Notre Dame 11-2
13) Baylor 11-3
14) Auburn 9-4
15) Iowa 10-3
16) Utah 11-3
17) Memphis 12-2
18) Michigan 9-4
19) Appalachian State 13-1
20) Navy 11-2
21) Cincinnati 11-3
22) Air Force 11-2
23) Boise State 12-2
24) Central Florida 10-3
25) Texas 8-5

Here is the final AP poll college football top 25 for the 2019 season. The fixed final AP poll top 25 follows the article below.

But before I get to correcting the AP poll's errors, let me make one thing clear. This is not about what I would personally prefer to see in the rankings. I myself would be inclined to rank Minnesota behind Wisconsin, and Penn State behind Minnesota, and I would definitely put Texas A&M, Oklahoma State, and Kansas State in my own top 25, just as a few examples. But in these cases, the choices the AP poll made are logically viable, if barely. This is about fixing only those AP poll choices that are not logically valid or fair.

For the third year in a row, most of the following fixes to the final AP poll are simply accounting for head-to-head results. AP poll voters have a very hard time accounting for much at all beyond straight record and the result of the last game played. They tend to disregard strength of schedule entirely, and that leads to irrational rankings.

For example, let's say a team goes 9-4, losing to #3, #8, #9, and #11, and defeating #12 and #15. Where should that team be ranked? Obviously between #11 and #12, behind all of the teams that beat them, and ahead of all the teams they defeated. But the AP poll ranks that team #18, behind two teams they beat!

Completely senseless. But we'll fix it below...

Louisiana State quarterback Joe Burrow in the national championship game

My ms-painting of Heisman Trophy winning quarterback Joe Burrow leading Louisiana State to a 42-25 victory over Clemson in the national championship game. Burrow had a season for the ages, including a record 60 touchdown passes.

Auburn > Oregon

I predicted and addressed most of the problems with this AP poll top 25 in an article I published not long after the New Year's Day bowls. The first issue I wrote about there was Oregon and Auburn. These 2 teams stand out as problems in the final AP ranking because #5 Oregon has 2 losses to lower-ranked teams, so they are overrated compared to the teams just behind them, and #14 Auburn has 2 wins over higher-ranked teams, so they are underrated compared to the teams just ahead of them. Luckily, both of these problems can be addressed with one simple fix. Auburn actually played and defeated Oregon, 27-21 in the season opener for both teams. If we move Auburn ahead of Oregon, where they belong, then the rating of both teams will make a lot more sense, as Oregon will have a more typical one upset loss (to unranked Arizona State), and Auburn will have one upset win (over Alabama), balanced out by one upset loss (to Minnesota in their bowl game).

Sure, Auburn is 9-4 and Oregon 12-2, but Auburn played 6 top 10 teams and Oregon played none. So let's compare apples to apples and oranges to oranges. Auburn was 2-4 against top 10 teams, Oregon 0-0. Auburn was 7-0 against all other teams, Oregon 12-2. So you see, Auburn actually had the better relevant record. AP poll voters are just punishing Auburn for playing a vastly tougher schedule, which is unfortunately the sort of thing they do every single year.

Auburn needs to be moved ahead of Oregon. Where do we put them? The average ranking of #5 Oregon and #14 Auburn is 9.5, so we'll place these teams behind #9 Penn State and ahead of #10 Minnesota. I don't like this ranking, and I would obviously not rank them this way myself, but I think it will just barely work. Still, it's worth looking at my problems here in more detail...

Penn State, Minnesota, and Wisconsin

First of all, the AP poll has rated #9 Penn State, #10 Minnesota, and #11 Wisconsin in the opposite order of how I would rate these teams. Because Wisconsin demolished Minnesota 38-17 in their regular season finale, thereby winning their division of the Big 10 over Minnesota, and because Minnesota beat Penn State 31-26, I would order these teams like so: Wisconsin > Minnesota > Penn State. But while I think this is a far better ordering for these teams, I do think that the AP poll's ordering of these teams, while weaker, is logically viable.

Penn State's loss to Minnesota was close (31-26) and came on the road, so it was not decisive. Penn State beat Iowa, who beat Minnesota. Penn State beat 3 ranked teams and 7 bowl qualifiers, while Minnesota beat 2 ranked teams and only 4 bowl qualifiers. And Penn State vastly outperformed Minnesota on the season, with 1 close win over an unranked team to Minnesota's 4 (note: Indiana scored a late touchdown to make their game with Penn State look closer than it was). Throw in Minnesota's 21 point loss to Wisconsin in their regular season finale, and the performance gap widens. Due to all this, Penn State can be viably rated ahead of Minnesota despite the head-to-head result (note that none of this is true for Oregon compared to Auburn, which is why dismissing Oregon's head-to-head loss to Auburn is not logically viable).

And with Penn State and Auburn both ranked higher than Minnesota, that gives Minnesota 2 wins over higher-ranked teams, which outweighs their one head-to-head loss to Wisconsin, and as such, they can be logically rated higher than Wisconsin. So you see, moving Auburn ahead of Oregon not only fixes the problems of an overrated Oregon and an underrated Auburn, it also fixes the issue of Minnesota being rated higher than Wisconsin.

Penn State vs. Auburn

I have a problem with Penn State being ranked ahead of Auburn due to Auburn's win over Alabama, who is ranked higher than Penn State. I would rate Auburn higher myself, but I can give this one to the AP poll. PSU, again, beat 3 ranked teams and 7 bowl qualifiers, while Auburn beat 2 ranked teams and 6 bowl qualifiers, so there's an edge to PSU. And Auburn's one poor performance, a 20-14 win over 4-8 Mississippi, was worse than PSU's one poor performance, a 17-10 win over 8-5 Pittsburgh. And while both teams took one upset loss, to Minnesota, PSU's loss came on the road, while Auburn's came on a neutral field, and in their bowl game, giving it greater weight.

Oregon vs. Minnesota

I also have a problem with Oregon being ranked ahead of Minnesota in this way. Note that if I were ranking these teams, it would look like this: Auburn > Oregon > Wisconsin > Minnesota > Penn State. This would follow the victory chain and would present no logical problems. Minnesota did beat Auburn in their bowl game, but Auburn has a win over a higher-rated team (Alabama), and Minnesota has an upset loss (to Iowa).

However, once you move Penn State up ahead of all these teams, giving Minnesota 2 wins over higher-rated teams, and when you move Minnesota ahead of Wisconsin due to those big wins, then it seems like you should move Minnesota ahead of Oregon as well. In this position, Minnesota has 2 upset losses (Iowa and Wisconsin) and 2 upset wins (Auburn and Penn State), which balance each other out. Oregon, on the other hand, has 1 upset loss (to unranked Arizona State) and no upset wins. So Minnesota has a relative record that is effectively 1 game better than that of Oregon for the season.

But I can give this one to the AP poll for similar reasons that Penn State can be rated higher than Minnesota. Oregon and Minnesota both beat 2 ranked teams, but Oregon beat 7 bowl qualifiers and Minnesota only beat 4. Minnesota's schedule really was atrocious. And Oregon performed better, posting 2 close wins over unranked teams to Minnesota's 4. Oregon also beat Wisconsin in the Rose Bowl, and Wisconsin trounced Minnesota 38-17 in their regular season finale.

My only remaining problem with all this is that Wisconsin also played a tougher schedule than Minnesota and performed far better. So why are we using schedule and performance to disregard Minnesota's better relevant record in one case (Oregon) but not in the other (Wisconsin)? This is a sticking point that makes the AP poll's choices rather weak to me, but I suppose we can see it this way: Minnesota vs. Wisconsin and Minnesota vs. Oregon are both very close cases, but Oregon fared better because they were slightly better than Wisconsin, as evidenced by their 1 point win over Wisconsin in the Rose Bowl.

Our First Fix

So again, our first fix is to move Auburn ahead of Oregon, and to place both teams behind Penn State. That results in the following ratings shift: Florida #5, Oklahoma #6, Alabama #7, Penn State #8, Auburn #9, Oregon #10, Minnesota #11, Wisconsin #12, Notre Dame #13, and Baylor #14.

Michigan > Notre Dame and Iowa

This one is a no-brainer. Michigan is the team in the example at the top of this article. They lost to #3, #8, #9, and #11, and they defeated #12 and #15. Rating them should be as simple as it gets, since they can easily be slotted right between all the teams that beat them and all the teams they beat. This is a head-to-head issue, and in this case, the AP poll has ignored not one, but two head-to-head results. 9-4 Michigan (#18) defeated 11-2 Notre Dame (now #13) and 10-3 Iowa (#15), and they should definitely be rated ahead of both teams.

Michigan destroyed Notre Dame 45-14. I would call that decisive. All 4 of Michigan's losses came to teams ranked higher than Notre Dame. And Michigan defeated 2 ranked teams, Notre Dame only 1 (Navy). There is just no real debate to be had here.

Baylor

#18 Michigan should be moved ahead of now-#13 Notre Dame and #15 Iowa, and the average rating of the 3 teams would place them behind #14 Baylor, where Iowa is now. However, Baylor does not compare well to any of these 3 teams, and so I propose moving them all ahead of Baylor, where Notre Dame is now. Baylor beat just 1 rated team, #25 Texas, they beat just 4 bowl qualifiers, and their performances were too often very poor. They posted 5 close wins over unranked teams, 4 of those against losing teams. Let's cut to the chase and compare them with Iowa in these regards. Iowa also had just 1 win over a rated opponent, but it came over a higher-rated team (now-#11 Minnesota), so it was much better than Baylor's win over #25 Texas. And because Minnesota is rated higher than Baylor, Iowa has a better relevant record than Baylor does. Iowa's 5 wins over bowl qualifiers isn't all that impressive, but it's better than Baylor's 4. And Iowa posted just 3 close wins, 2 over losing teams. Iowa outdoes Baylor in every regard, so there is no reason for Baylor to be ranked ahead of Iowa.

Notre Dame vs. Iowa

That leaves one issue to address. Iowa's win over a higher-rated team (now-#11 Minnesota) gives them a better relevant record than Notre Dame has as well, and for that reason I would rank Iowa higher than Notre Dame. However, this is another one I can give to the AP poll. Notre Dame's one win over a rated team came 52-20 over #20 Navy, and while #11 Minnesota is unquestionably a better opponent than Navy, it is not at all clear that beating the #11 team by 4 points is better than beating the #20 team by 32 points. If we see those results as equitable, which I think is fair, then Iowa and Notre Dame look pretty much the same. Both posted 3 poor performances, but Notre Dame's last poor performance was November 2nd (a 21-20 home win over unranked Virginia Tech), while Iowa posted an even worse result in their regular season finale (27-24 at 5-7 Nebraska). On the basis of that tiny edge in timing, I'll let the AP poll keep Notre Dame in front of Iowa.

Our Second Fix

So again, our second fix is to move Michigan ahead of both Notre Dame and Iowa, and to place these 3 teams behind Wisconsin (who beat Michigan) and ahead of Baylor. That results in the following ratings shift: Michigan #13, Notre Dame #14, Iowa #15, Baylor #16, Utah #17, and Memphis #18.

Oklahoma

This is the first issue that was not covered in my January 3rd article, as I did not foresee this issue until I was carefully going over the AP poll's final top 25. At first glance, 12-2 Oklahoma (originally rated #7 in the final poll) appears to be a solid top 10 team, with 3 wins over rated teams and the typical 1 upset loss. However, as I compared the Sooners to the teams behind them, one by one they fell further and further down the list, and I could not find any good reason to let the AP poll have its way.

Let's start by comparing 12-2 Oklahoma to 11-2 Alabama, who was ranked right behind them. Alabama's upset loss came to Auburn (now ranked #9), while Oklahoma's upset loss came to unranked 8-5 Kansas State, so advantage Alabama there. Alabama's other loss came by 5 points to #1 LSU, who destroyed Oklahoma 63-28. Alabama's performance this season was otherwise spectacular, as the closest anyone got to them was 19 points. Seven teams came closer to Oklahoma than that (not including the 2 teams that beat them)! So this comparison is a rout in Alabama's favor.

Oklahoma has a better case against 11-2 Penn State, who was ranked right behind Alabama. PSU's upset loss came to Minnesota (now #11), so this is again better than Oklahoma's upset loss to unranked Kansas State. Penn State did not perform as well as Alabama, but they still outperformed Oklahoma. PSU posted 3 close wins, Oklahoma 5. Their schedules were comparable, as they both beat 3 ranked teams, but PSU beat more bowl qualifiers. Furthermore, the Big 10 was better than the Big 12 this season, going 38-13 (.745) against nonconference opponents, the Big 12 just 25-11 (.694). There's just nothing to point to in Oklahoma's favor here, so no good reason to rate them higher than Penn State.

Next is 9-4 Auburn, but despite Oklahoma's advantage in straight record here, this comparison goes worse for them than the PSU comparison. Auburn has a win over Alabama, which we have already established to be a better team than Oklahoma and deserving of a higher rating. So Auburn has a better relevant record than Oklahoma does. And similar to Alabama, Auburn only lost by 3 points to #1 LSU, who, again, trashed Oklahoma 63-28.

Next is 12-2 Oregon, who also performed better than Oklahoma did, and against a tougher schedule. I won't get into the details this time, because AP poll voters originally rated Oregon #5 and Oklahoma #7, so now that we are down to comparing Oklahoma with Oregon, we already know which one the poll voters prefer. Since I agree, we'll move on.

Out of the Top 10

Behind Oregon are a bunch of teams with better relevant records than Oklahoma has. Minnesota has 2 upset losses (Wisconsin and Iowa), but those are balanced out with 2 upset wins (Penn State and Auburn). Oklahoma's upset loss to unranked Kansas State, however, is not balanced out with any upset wins, so they are effectively 1 game worse than Minnesota for the season. Similarly, Wisconsin has 1 upset loss (Illinois) and 1 upset win (Minnesota), and Michigan, Notre Dame, and Iowa did not suffer any upset losses at all.

As we did with Baylor above, let's cut to the chase and compare 12-2 Oklahoma with 10-3 Iowa. First of all, remember that the Big 10 was better than the Big 12 this season (.745 against nonconference opponents to .694), so it shouldn't come as a surprise that a 10-3 Big 10 team would be better than a 12-2 Big 12 team. Oklahoma has 3 wins over rated teams, Iowa just 1, but that one for Iowa came over a higher-rated team, Minnesota (now #11), and another team Iowa defeated will be ranked in the fixed poll (Southern Cal, as covered below). As such, this edge for Oklahoma doesn't come close to making up for the fact that Oklahoma took an upset loss to an unranked team, while Iowa took no upset losses at all. As for performance, Oklahoma posted 4 poor performances, Iowa 3, and Iowa's 25 point win over Southern Cal (will be ranked #23 in fixed poll) is far better than Oklahoma's 7 point win over Texas (#25).

So we're going to drop Oklahoma all the way back behind Iowa, just in front of Baylor. Reinforcing this ranking is the fact that Oklahoma only beat Baylor by 3 points the first time, and in overtime the second, indicating that they are only just barely better than Baylor, and ought to be ranked just in front of Baylor. Oklahoma drops to #15, and all the teams that had been ranked between Oklahoma and Baylor move up 1 slot.

Texas > Utah

Here is another head-to-head issue I addressed in my previous article. Texas may be 8-5 and Utah 11-3, but Texas embarrassed Utah 38-10 in the Alamo Bowl, and they should certainly be rated ahead of Utah. It just doesn't get any clearer than 38-10, and it came in a bowl game, giving it greater emphasis. Texas did take 2 upset losses to unranked teams (TCU and Iowa State), but Utah took 1 (Southern Cal), so the bowl result makes them effectively even (Texas' other 3 losses all came to teams ranked higher than Utah, so those losses should be irrelevant when comparing Texas to Utah). The head-to-head tiebreaker goes to Texas, and again, it was a resounding score in a bowl game.

Texas should be moved ahead of Utah, and the average of the original ratings of Texas (#25) and Utah (#16) would place both teams behind #20 Navy and ahead of #21 Cincinnati, which works fine. Cincinnati didn't take any upset losses, while Utah took the one to unranked Southern Cal, but USC is better than any team Cincinnati was able to beat themselves (USC will be ranked in the fixed poll, while Cinci victim Central Florida will not be, as covered below). In the fixed poll, Utah will have a win over a ranked team (Washington, also covered below) and Cincinnati will not.

We have another big ratings shift: Memphis #17, Appalachian State #18, Navy #19, Texas #20, Utah #21, Cincinnati #22, Air Force #23, Boise State #24, and Central Florida #25.

Navy and Appalachian State

This is another issue I addressed in my previous article, albeit briefly. I'll just repeat what I wrote there: 13-1 Appalachian State took an upset loss to an unranked team (7-6 Georgia Southern), and they should therefore be rated behind 11-2 Navy, who took no upset losses. It need hardly be said that the AAC was a far better conference than the Sun Belt, and Navy accomplished far more than Appalachian State did this season, including wins over 11-2 Air Force (now #23), 10-3 Southern Methodist, and 8-5 Kansas State.

Appalachian State being ranked ahead of Texas works fine, since Texas took 2 upset losses to unranked teams.

Switch 'em. Navy #18, Appalachian State #19.

Air Force, Boise State, Washington, and Southern Cal

Also covered in my previous article, I think that these teams should be ranked like so: 8-5 Southern Cal > 8-5 Washington > 12-2 Boise State > 11-2 Air Force. The AP poll, unfortunately, has rated them in the reverse order, with Air Force #22 (original rating), Boise State #23, Washington #28 (in Others Receiving Votes section), and Southern Cal #30. The Big issue here is Washington's 38-7 obliteration of Boise State in the Las Vegas Bowl. I just don't see how you can simply ignore such a clear and decisive result. Then there's the fact that Boise State beat Air Force 30-19, and easily won the Mountain West conference.

As I said in that previous article, I would go even further myself, and rate these teams into this long victory chain: 8-5 Southern Cal > 8-5 Arizona State > 8-5 Cal > 8-5 Washington > 12-2 Boise State > 11-2 Air Force. That's because USC beat both ASU and Cal, ASU beat Cal, Cal beat Washington, Washington destroyed Boise State in a bowl game, and Boise State beat Air Force. However, I do think that Arizona State and California can be removed from this chain and validly dismissed. ASU took a huge 3 upset losses (Boise State only took 1, and Air Force none), and their upset win over Oregon only makes up for one of those. And Cal's 20-19 win at Washington was their only win over a winning team. They otherwise performed very poorly against a very weak schedule. Washington beat 4 winning teams and performed far better, so Washington can be validly rated ahead of Cal despite the head-to-head result.

But Southern Cal is another story. Washington did beat Southern Cal this season, and both teams finished 8-5, but there is simply no getting around the fact that Southern Cal should be rated ahead of Washington. That is because USC beat four teams that beat Washington this year: Utah, Cal, Stanford, and Colorado. Washington beat BYU, who beat USC, and their head-to-head win over the Trojans makes up for another 1 of those games, but that's it, leaving Washington effectively 2 games worse than Southern Cal for the season.

So USC > Washington > Boise State > Air Force. Where do we put them? The average of the original AP ratings of these teams is 25.75, which would place them just ahead of #26 Texas A&M. This is a tough one, because I would put 8-5 Texas A&M into my own top 25, and fairly highly. All 5 of their losses have come to teams that are currently ranked in the top 9. However, 11-2 Air Force has not taken any upset losses either, and I think that rating Air Force ahead of Texas A&M is just doable. Air Force hasn't beaten anyone that is all that good, and they have only beaten 4 bowl qualifiers, but Texas A&M has only beaten 2 bowl qualifiers, which is quite pathetic.

We have another ratings shift: Central Florida #23, Southern Cal #24, Washington #25, Boise State #26, and Air Force #27. So Boise State and Air Force are out of the rankings... for the moment.

Central Florida

For me, the biggest surprise when the final AP poll was published was 10-3 Central Florida at #24. They were not even close to rated in the final regular season AP poll, clocking in with just 5 poll points. Washington was right behind them with 3 poll points. Then UCF beat a worthless Marshall (8-5) 48-25 in the Gasparilla Bowl, while Washington beat #18 Boise State 38-7 in the Las Vegas Bowl, and somehow, UCF jetted all the way into the top 25, while Washington only moved to #28. I am still mystified by this.

With 2 losses to unranked teams (Pittsburgh and Tulsa), and not much to hang their hats on in terms of wins, Central Florida is a rather dubious inclusion in the top 25. They did beat 11-3 Florida Atlantic, who sits just outside the top 25 (currently #29). However, that pales next to Southern Cal's best win, which came over a higher-ranked team (Utah, now #21). That gives USC a better relevant record than UCF has, and Southern Cal also defeated 2 teams who were nearly rated (8-5 Arizona State and 8-5 California).

Central Florida compares far better with Washington, as the two have the same relevant record. These two look practically the same in performance too, and I considered leaving UCF ranked higher than Washington. There was just one nagging problem. In the original AP poll, Air Force and Boise State were both rated higher than Central Florida, so once it's clear that Southern Cal should be rated ahead of UCF, I might as well drop UCF all the way back behind Air Force, as that would be closer to what AP poll voters wanted. After all, if I put Central Florida ahead of Washington, then UCF will be ranked ahead of 2 teams the AP poll ranked above them, but if I put them behind Air Force, they'll be ranked behind just one team that the AP poll ranked behind them. And there's certainly no logical problem with rating Air Force ahead of Central Florida, since Air Force took no upset losses and UCF took 2.

So we'll send Central Florida back behind Air Force, which is to say, out of the rankings. Southern Cal moves up to #23, Washington to #24, and Boise State moves back into the rankings at #25.

Fixed AP Poll Top 25

Two teams fall out of this fixed AP poll top 25, 11-2 Air Force (#22) and 10-3 Central Florida (#24). They are replaced by 8-5 Southern Cal (now #23) and 8-5 Washington (#24). The 2 teams that fall out totaled 2 losses to unranked teams, and they did not defeat any AP-rated teams. Discounting their game against each other, the 2 teams that replace them combined for a much worse 4 losses to unranked teams, but on the bright side, they notched a much better 2 wins against AP-rated teams. So that's a wash this year.

And here it is, your fixed final AP poll for the 2019 season, now logically coherent.

1) Louisiana State 15-0	--
2) Clemson 14-1	--
3) Ohio State 13-1	--
4) Georgia 12-2	--
5) Florida 11-2	+1
6) Alabama 11-2	+2
7) Penn State 11-2	+2
8) Auburn 9-4	+6
9) Oregon 12-2	-4
10) Minnesota 11-2	--
11) Wisconsin 10-4	--
12) Michigan 9-4	+6
13) Notre Dame 11-2	-1
14) Iowa 10-3	+1
15) Oklahoma 12-2	-8
16) Baylor 11-3	-3
17) Memphis 12-2	--
18) Navy 11-2	+2
19) Appalachian State 13-1	--
20) Texas 8-5	+5
21) Utah 11-3	-5
22) Cincinnati 11-3	-1
23) Southern Cal 8-5	IN
24) Washington 8-5	IN
25) Boise State 12-2	-2

OUT: #22 Air Force 11-2
#24 Central Florida 10-3

Fixed AP Polls
Home