An In Depth Analysis of College Football Polling

You all know my ire for the BCS. I can’t stand the system at all, but it did do one good thing for the sport- it sorted out the mess of the polls. Many people don’t know what it means or ever meant to be the champions of college football. Come to think of it, the BCS muddled that a bit too, but I digress. Before 1998, the champion was whoever was voted at #1 after the bowl games. Before 1974, the coaches poll didn’t even come out again after the bowls, so it was whoever won the AP Poll after the bowl games. Pretty archaic way of choosing a champion. Yet these polls persist in playing a role in modern college football. Let’s take an in depth look at why they are the most outdated, illegitimate way of choosing a champion there is.

I might add, the BCS only requires the coaches to vote the winner of the BCS championship game as number one. Kyle Whittingham didn’t even do that last year, putting his undefeated Utes there instead. But still, at least there is almost always consensus there. Not so in the AP Poll. It can still get all over the board. So in theory, you could have had Utah win one poll last year and Florida win another. That would make two national champions according to what it means to be national champion at the moment.

So let us look at the polls that matter, the AP and USA Today. The AP Poll this year used 60 voters. Their votes are public (although dang near impossible to find) and are released each week. There is typically some consistency after the preseason polls. You have the occasional outlier like Craig “I’m the world’s biggest idiot” James. But by and large once they see how everyone else voted in the preseason polls, they quickly align and put teams in pretty much the same place. The USA Today poll takes a sampling of 58 of the 120 FBS coaches and gives them a vote. They are evenly distributed across conferences, which is a good thing. I believe these polls are also public, although they tried to change that recently. Not sure how that adds transparency or accountability to the sport, but hey, that’s the BCS for you.

Now here is the pathetic thing about polls. Even after the New England Patriots lost the Super Bowl to the New York Giants, if they used polls, the Pats would have been NFL champs. No way you can say the team with only one loss in 18 or 19 games is worse than the team that just beat them and had way more losses. Not to mention the Pats beat the Giants already in the regular season, so, Pats win the title.

Fair? Just? Of course not! Both teams EARNED a spot in the Super Bowl by winning on the field, not winning the votes of random fans and media folks. With polls, the Giants would never have been in the Super Bowl at all. The top two would have just played for it to bypass the inconvenience of a playoff. And so it goes with any sport. The regular season is to jockey for playoff position. Then the playoff is for who gets to compete in the championship. And the championship game determines who wins the title for the league that year. It’s pretty simple. Every sport does this, even college football in its other three divisions. But for some reason, the FBS relies heavily on human voters to pick the best teams in the country. Let’s look at one MAJOR flaw in this system.

First we will look at the AP Poll for the 2009 season. I will put the preseason poll and the final poll side by side for comparison. Teams in red in the preseason poll were not ranked at the end of the season. Teams in blue in the final poll didn’t start the season ranked. Let’s see the incredible fallibility of preseason polls:


Preseason AP Poll
1 Florida (58)
2 Texas (2)
3 Oklahoma
4 USC
5 Alabama
6 Ohio State
7 Virginia Tech
8 Mississippi
9 Oklahoma State
9 Penn State
11 LSU
12 California
13 Georgia
14 Boise State
15 Georgia Tech
16 Oregon
17 TCU
18 Florida State
19 Utah
20 Brigham Young
21 North Carolina
22 Iowa
23 Notre Dame
24 Nebraska
25 Kansas


Final AP Poll
1 Alabama (60)
2 Texas
3 Florida
4 Boise State
5 Ohio State
6 TCU
7 Iowa
8 Cincinnati
9 Penn State
10 Virginia Tech
11 Oregon
12 Brigham Young
13 Georgia Tech
14 Nebraska
15 Pittsburgh
16 Wisconsin
17 LSU
18 Utah
19 Miami (FL)
20 Mississippi
21 Texas Tech
22 USC
23 Central Michigan
24 Clemson
25 West Virginia

As we can see, 28% of the teams ranked in the preseason poll were not ranked in the final poll, and 32% of the teams ranked in the final poll did not start the season ranked. In fact, there were 29 teams not ranked in the preseason poll that still got votes. Only 15 teams in the final poll got votes but didn’t make the top 25. So this begs the question- why are there preseason polls? Who knows, I just know they are useless. To further demonstrate, let’s look at exhibit B, the USA Today poll:


Preseason USA Today Poll
1 Florida (53)
2 Texas (4)
3 Oklahoma (1)
4 USC (1)
5 Alabama
6 Ohio State
7 Virginia Tech
8 Penn State
9 LSU
10 Mississippi
11 Oklahoma State
12 California
13 Georgia
14 Oregon
15 Georgia Tech
16 Boise State
17 TCU
18 Utah
19 Florida State
20 North Carolina
21 Iowa
22 Nebraska
23 Notre Dame
24 Brigham Young
25 Oregon State


Final USA Today Poll
1 Alabama (58)
2 Texas
3 Florida
4 Boise State
5 Ohio State
6 TCU
7 Iowa
8 Penn State
9 Cincinnati
10 Virginia Tech
11 Oregon
12 Brigham Young
13 Georgia Tech
14 Nebraska
15 Pittsburgh
16 Wisconsin
17 LSU
18 Utah
19 Miami (FL)
20 USC
21 Mississippi
22 West Virginia
23 Texas Tech
24 Central Michigan
25 Oklahoma State

So what gives? In both instances the pollsters only “get it right” (I use the term loosely) about 2/3 of the time. I present this not just to show how irrelevant preseason polls are, but to show the illegitimacy of the voters. They CLEARLY don’t know what they are doing. They try to use their bias to get their teams ahead and still stay close to the computers, while the computer formulas are COMPLETELY bias towards the “big” schools. Oh, you think the computer rankings are good and fair because they are computers? Don’t fool yourself, I know statistics, and I know how easy it is to tweak one tiny part of the algorithm to get the outcome you want.

Look at the most widely heralded and accepted computer rankings, the Jeff Sagarin rankings. His final rankings put TCU one spot ahead of Boise State. Didn’t the Broncos JUST BEAT the Horned Frogs? They put Florida ahead of Texas. Didn’t Texas ALMOST BEAT Alabama, after the Tide destroyed the Gators? The Texas did it with a backup freshman quarterback! It gets better, Virginia Tech (10-3) ended up one spot ahead of Cincinnati (12-1). The Hokies also finished ahead of 11-3 Georgia Tech, a team the Hokies lost to in the regular season. 8-5 bottom dwellers Oklahoma and Arkansas finished 17 and 18 while 10-3 Wisconsin ended up at 23. Wisconsin beat Miami (FL), a team that beat Oklahoma. 12-2 Central Michigan finished at an astounding 42, while sitting at 36 was 7-6 Tennessee.

So you have human polls and computer polls that both are terrible. Both have bias. Both favor “big time” programs over the little guys. Both keep our archaic system in place. And neither looks like it will change any time soon.

We all know the old adage “If you can’t beat them, join them…then beat them.” Well, we can’t beat the polls. We can’t beat the powers that be. We have to join them…then beat them. Which is why I am pleased to announce something very important. Yours truly is going to create his own computer rankings. Go ahead…laugh it up, I don’t care. I am a professional in statistics and I know what I’m doing. My plan is to use a series of Chi Square tests with different data. This tells me if the observed values differ significantly from the expected values.

Stay tuned for more details on this to come. I’ll use next season as my test run on these rankings. I’ll get it figured out. Either way, I’m going to show them how to find real rankings without bias. Hopefully one day it can set the precedent to make the polls completely irrelevant and disappear.


Comments