Tip Top 25 in helmets, smaller
                                                    Home

Fixing the Final 2005 AP College Football Poll

1) Texas 13-0
2) Southern Cal 12-1
3) Penn State 11-1
4) Ohio State 10-2
5) West Virginia 11-1
6) Louisiana State 11-2
7) Virginia Tech 11-2
8) Alabama 10-2
9) Notre Dame 9-3
10) Georgia 10-3
11) Texas Christian 11-1
12) Florida 9-3
13) Oregon 10-2
14) Auburn 9-3
15) Wisconsin 10-3
16) UCLA 10-2
17) Miami (FL) 9-3
18) Boston College 9-3
19) Louisville 9-3
20) Texas Tech 9-3
21) Clemson 8-4
22) Oklahoma 8-4
23) Florida State 8-5
24) Nebraska 8-4
25) California 8-4


Here is the final 2005 AP college football top 25. You can access all of these teams' full schedules at the College Football Data Warehouse (amongst a number of other places). The fixed final AP top 25 follows the article below. 

The first thing I notice here is SEC champion Georgia (10-3) being ranked four slots behind 11-2 Louisiana State, whom Georgia defeated resoundingly in the SEC title game, 34-14. Has an injustice been dealt to the Bulldogs?

I myself would rank Georgia ahead of LSU, as their head-to-head win was very decisive, came at the end of the season when the stakes were at their highest, and Georgia's bowl loss was by 3 points to a team ranked higher than LSU (West Virginia).

However, AP poll voters do have a case for rating LSU higher this year...

Vince Young winning touchdown against USC

My painting of the 2005 season's play of the yearVince Young scoring the winning touchdown from 9 yards out on 4th and 5 with 19 seconds left. Thus ended Southern Cal's pursuit of a "threepeat," as well as talk of USC 2005 being the greatest team of all time.

Georgia

AP voters moved LSU ahead of Georgia in the final poll because LSU demolished #17 Miami in the Peach Bowl 40-3, while Georgia lost to #5 West Virginia in the Sugar Bowl 38-35. LSU was very impressive in their bowl game, no doubt, but those results alone are not nearly enough for LSU to be logically rated ahead of Georgia, considering the 34-14 beatdown Georgia put on LSU in the previous game.

But where AP voters have a case for rating LSU higher is the fact that LSU defeated a big four ranked opponents. On the other hand, Georgia's only win over a ranked opponent is that SEC title game win over LSU. They lost their other 3 games against ranked opponents, and that gives reason to perceive the SEC title game as an anomaly.

However, I fail to see the logic in rating 9-3 Notre Dame just in front of Georgia. Georgia's win over LSU is one more win over a ranked opponent than Notre Dame has. Georgia lost to 2 lower-rated teams, but those teams, #12 Florida (9-3) and #14 Auburn (9-3), are far better than anyone Notre Dame defeated, so there is no reason to believe that Notre Dame would have fared better against them. And Notre Dame lost to 5-6 Michigan State!

So we'll move Georgia to #9 and drop Notre Dame to #10.

Florida and Auburn

And while we're at it, why are Florida and Auburn ranked behind Notre Dame? Again, Notre Dame has no win over a ranked opponent and a loss to 5-6 Michigan State. Florida and Auburn both have better relevant records, and each has 2 wins over rated opponents. So let's move them both behind Georgia and ahead of Notre Dame. In the process, they also both jump over 11-1 Texas Christian and 10-2 Oregon. That's fine, because TCU and Oregon will both be dropping anyway (details below).

So now we have the following: Georgia #9, Florida #10, Auburn #11, Notre Dame #12, TCU #13, and Oregon #14. You may notice that Florida now sits right behind Georgia, a team Florida defeated 14-10, and they both have equal relevant records. So should Florida be moved ahead of Georgia?

Georgia vs. Florida

The problem is that the AP poll has Georgia ranked behind not only LSU, but Alabama as well. Those are valid choices, but it makes rating Georgia ahead of Florida problematic. Florida defeated Georgia, but lost to LSU and Alabama. So if the AP poll had rated Georgia ahead of Alabama, Florida's win over Georgia would not be a problem, since the Gators would be stuck behind Alabama. As I said earlier, I myself would rate Georgia ahead of LSU. But if I didn't, and if I also decided to rank Alabama ahead of Georgia, then I would definitely put Florida ahead of Georgia too.

However, as with the decision to rate LSU higher than Georgia, the AP poll does have a case for rating Georgia higher than Florida. But this time, instead of strength of schedule, the issue is performance. Georgia had a higher score differential than Florida against both rated and unrated opponents. More importantly, after you toss out all the similar performances for each team, Georgia is left with a 3-game advantage in terms of performance. Those 3 comparable games for each team all happened in the second half of the season, and are as follows:
  1. Florida's best win is their 14-10 win over now-#9 Georgia, while Georgia's is their 34-14 win over #6 LSU.
  2. Georgia's worst loss was an upset to now-#11 Auburn 31-30, while Florida was upset by unranked South Carolina 30-22.
  3. In their bowl games, Georgia lost to #5 West Virginia 38-35 (a #6-10 type performance), while Florida beat unranked Iowa 31-24 (a #20-25 type performance).
The performances in all the rest of their games equal out. But the performance difference above is wide enough to allow the AP poll to keep Georgia ahead of Florida.

Florida State

Finally, we come to a head-to-head result the AP poll ignored that it has no good case for ignoring. Florida State (8-5, #23) won at Boston College (9-3, #18) 28-17, and the only reason BC is rated 5 spots higher, of course, is that FSU had 5 losses and BC had 3. However, FSU played 6 rated opponents, and BC only played 3. So FSU played twice as many top 25 opponents, but they took less than twice as many losses.

Despite the fact that BC has a better straight record by 2 games, FSU actually has the better real or relevant record. Ignoring their game against each other for the moment, BC lost to one lower-ranked team, and defeated none that were higher-ranked. FSU was incredibly erratic, losing to 3 teams ranked lower than BC, and defeating 2 higher-ranked teams (one of which was #7 Virginia Tech in the ACC title game, making FSU the ACC champion). FSU's 2 upset wins offset 2 of their 3 upset losses, leaving them with 1 upset loss, just like BC has. Now add in the fact that FSU defeated BC, and FSU is effectively one game better than BC.

FSU just played a tougher schedule. Twice as tough, in fact. BC may have a slight performance advantage (I see them as equal performance-wise), but it is nowhere near enough to dismiss the head-to-head result here. Especially given that it was a decisive 28-17 road win.

The AP poll punishes teams for playing tough schedules and rewards teams for playing weak ones. And it shouldn't. So we'll fix this by moving Florida State ahead of Boston College. The only question is, where should we put them?

We'll have them meet in the middle of where they are ranked now, just behind #20 Texas Tech and ahead of #21 Clemson. That works perfectly, since Boston College defeated Clemson. Clemson did beat FSU, but FSU has the better relevant record and should be ranked ahead of Clemson anyway. The rankings now shift as follows: Louisville #18, Texas Tech #19, FSU #20, BC #21, Clemson #22, and Oklahoma #23.

Oregon

Here is a very interesting case. Oregon went into the Holiday Bowl at 10-1, while their opponent, Oklahoma, was 7-4. Oregon's only loss was to Southern Cal (#1 at the time), and they were rightly ranked in the top ten, at #6. Oklahoma was, of course, unranked. But Oklahoma won the game 17-14. Now, almost every time a 7-4 team is matched up with a 10-1 team in a bowl and wins, they should nevertheless be ranked lower in the final poll (assuming that both teams are from comparable conferences). However, this is not one of those times.

This is one of those times when just one"bad" game can and should have a drastic effect on a team's rating. It is a perfect example of a game that should make you completely reevaluate your preconceptions from before the game. The typical AP poll voter simply drops a team a few places for a loss, and raises a team for a win. So in a case like this, there was no way typical voters were going to drop their #6 team all the way behind a team they didn't even have in their top 25. But they should have.

Let's look at Oregon. They lost to the AP poll's #2 and #22 teams (USC and Oklahoma). And they defeated only one ranked team, #25 California. Is it not logical to think that they should be ranked #23 or #24? Between the teams they defeated and lost to? And the loss to Oklahoma was in a bowl game, giving it greater weight. And the win against #25 Cal was at home, in overtime, indicating that they should be ranked much closer to Cal than the AP poll had them anyway.

All four of Oklahoma's losses were to teams ranked in the AP poll's top 20. Oregon did not defeat any team ranked that highly, so there is no reason to assume that they would have done better against those opponents. Oklahoma defeated two ranked opponents to Oregon's one. And although Oregon scored much more on their weak opponents than Oklahoma did, the season performance of each team was actually equal: they each had 3 poor performances.

Forget 10-2 and 8-4: those win-loss records are mirages built on the fact that Oregon played 3 ranked opponents while Oklahoma played 6. Oregon has no case at all for being ranked higher than Oklahoma.

As we did with Florida State and Boston College, we'll move these two teams in between where they are ranked now. Since Oklahoma lost to Texas Tech, we'll place them behind the Red Raiders, resulting in the following ratings shift: Texas Tech #18, Oklahoma #19, Oregon #20, Florida State #21, Boston College #22, and Clemson #23. Also, Wisconsin, UCLA, Miami, and Louisville each move up one spot due to Oregon's drop.

Texas Christian

Texas Christian has a smashing 11-1 record, but it came against a very weak schedule, and TCU's performance against that schedule was too often poor. Their best game came in their opener, a 17-10 win over now-#19 Oklahoma, their only ranked opponent. Otherwise, they lost to 5-6 Southern Methodist, and they barely won against unranked opponents four times (twice in overtime, and twice by 3 points, including a 27-24 win over 7-5 Iowa State in their bowl game). I myself would drop TCU to just ahead of Oklahoma.

But a case can be made for TCU to remain ahead of 10-2 UCLA, so we can at least do that much for the AP poll voters. UCLA now sits 2 places behind TCU. Directly behind the Horned Frogs is 10-3 Wisconsin. Wisconsin has the same relevant record as TCU, but their performance was far superior, with only two of their 10 wins coming by a touchdown or less (compared to 5 for TCU), and two of those wins are over teams that will finish in the final fixed top 25 (compared to 1 for TCU). And Wisconsin showed up huge in their bowl game, upsetting now-#11 Auburn 24-10. While TCU barely beat 7-5 Iowa State in theirs.

So we'll just have them switch places: Wisconsin to #13 and TCU to #14.

Louisville

Louisville, at 9-3 and now #17, does not have a win against a ranked opponent, and in fact they do not have a win over any of the 17 teams in the "also receiving votes" section either. They lost to 6-6 South Florida 45-14. Clearly overrated. I myself would probably not have ranked them in the top 25.

But on the other hand, after their first 3 games, when they took that upset loss to South Florida and also had their only close win (over Kentucky in their opener), they beat up on all their weak opponents like a top 25 team should. And they lost at #5 West Virginia 46-44, a top ten performance. So they can be ranked. Just not this highly.

The original AP poll had Boston College just in front of Louisville, so we'll drop Louisville back behind BC, and in front of Clemson. Clemson had one win over a ranked opponent, but they also had 2 losses to unranked opponents, so their relevant record is the same as Louisville's. Due to Louisville's better performance, they can stay ahead of Clemson despite their very weak schedule.

Louisville drops to #22, BC moves up to #21, and of course all the teams that had been between Louisville and BC move up a spot.

Michigan

Here's a rarity: the AP poll respecting a head-to-head result that they shouldn't. That would be 8-4 Nebraska's 32-28 Alamo Bowl win over 7-5 Michigan. Obviously they respected it only because Nebraska is 8-4 and Michigan is 7-5. But Michigan actually has the better relevant record, and they were vastly better performance-wise.

Nebraska has 2 losses to unranked opponents, while Michigan has only one. Nebraska's bowl win over Michigan makes them even. However, Michigan's upset win over #3 Penn State (PSU's only loss!) moves Michigan effectively one game ahead of Nebraska. The only way you can see them as even is if you count the bowl game twice. And even then, Michigan's schedule and performance are so much better that they are not close. And Nebraska barely won that Alamo Bowl. It was not at all a decisive outcome.

I covered a similar situation when discussing the Rose Bowl between Southern Cal and Michigan in my Fixing the 2006 AP Poll article. But in that case, Southern Cal beat Michigan by 2 touchdowns, played and defeated more rated opponents, and performed very strongly on the season. Nebraska has no such advantages over Michigan in 2005.

Nebraska beat unranked opponents Pittsburgh 7-6, Iowa State 27-20 in overtime, and Kansas State 27-25, all at home. They lost at unranked Missouri 41-24 and at unranked Kansas 40-15. Those are terrible performances. The only thing Nebraska has to say for itself is that 4-point Alamo Bowl win over Michigan.

Michigan had 2 close wins over unrated opponents, one less than Nebraska, and furthermore, both of those games were on the road, and one was against a rival (Michigan State), and another was against a team just outside the top 25 (Iowa). Michigan's upset loss was by 3 points, and in fact, none of Michigan's losses were by more than 7 points, whereas Nebraska was destroyed twice by unrated opponents. Michigan lost to #4 Ohio State by only 4 points, and as I mentioned, they defeated #3 Penn State, which is so much better than Nebraska's best win that they are not comparable at all. Finally, Michigan also defeated 2 teams in the "also receiving votes" section, and Nebraska defeated none.

This is no contest, even when we give the Alamo Bowl result greater weight. So we'll move Michigan into the top 25 at #24, drop Nebraska to #25, and California thus falls out. Which is fine, since Cal didn't beat anyone of value anyway. Michigan can stay behind #23 Clemson due to their upset loss in a bowl game.

As for Nebraska, getting their biggest win in their bowl game was good timing, because the rest of their season was exceedingly poor, nowhere close to top 25 material.

Fixed AP Top 25

So, in the end, one team comes in (Michigan) and one falls out (Cal). Both teams had one loss to an unranked opponent, but only Michigan had a win against a ranked opponent, so this is, as always, an upgrade. On top of that, all the shifting around has greatly increased the structural integrity of the entire top 25, and it makes much more sense now.

It is interesting to see all those SEC teams bunched together, and it makes sense, because there wasn't much difference between the top 5 teams in the SEC in 2005. In fact, you could validly rank them in a lot of different ways relative to each other. I myself would probably rank them as follows: Georgia>LSU>Auburn>Alabama>Florida. But the AP poll's ranking of them is valid too.


1) Texas 13-0--
2) Southern Cal 12-1--
3) Penn State 11-1--
4) Ohio State 10-2--
5) West Virginia 11-1--
6) Louisiana State 11-2--
7) Virginia Tech 11-2--
8) Alabama 10-2--
9) Georgia 10-3+1
10) Florida 9-3+2
11) Auburn 9-3+3
12) Notre Dame 9-3-3
13) Wisconsin 10-3+2
14) Texas Christian 11-1-3
15) UCLA 10-2+1
16) Miami-Florida 9-3+1
17) Texas Tech 9-3+3
18) Oklahoma 8-4+4
19) Oregon 10-2-6
20) Florida State 8-5+3
21) Boston College 9-3-3
22) Louisville 9-3-3
23) Clemson 8-4-2
24) Michigan 7-5IN
25) Nebraska 8-4-1

OUT: #25 California 8-4

Home