FEI Is My New Favorite Poll

This article was originally posted on Tar Heel Mania, but I like the FEI system so much I’m going to post it here.

Many thanks to Tomahawk Nation for tipping me off to this.

As someone who majored in a hard science in college, I should probably be more knowledgeable about advanced football statistics. Alas, I have been pretty faithful to the box score up until now. But then I was tipped off to Brian Fremeau’s Efficiency Rankings for college football (FEI for short). It’s a very interesting formula, which you can read about in detail at Brian’s website and in Football Outsiders articles such as this one.

In essence, the rankings are based on the principle that a team’s efficiency in each offensive and defensive drive is a better reflection of a team’s quality than simple statistics accumulated over 60 minutes of play. Like some human and computer polls, the quality of opponents and how well a team plays against good teams. Unlike most human polls, the system rewards teams that play well against good competition whether they win or lose.

Fremeau’s Game Efficiency formula is fairly simple, involving margin of victory (MOV) and competitive possessions (CP, meaning the number of possessions in the game before garbage time, or the opposing team has fewer remaining possessions than is required for a comeback). The formula is as follows:

Game eff. = (MOV / 7) / (CP / 2)

Based on this rubric, one can get a pretty clear idea of how games stack up against one another. Let’s use three examples. First, Florida’s frighteningly efficient 63-5 victory over Kentucky:

Game eff = (58 / 7) / (16 / 2) = 8.2857 / 8 = 1.0357

Pretty scary, right? Conversely, here’s Auburn’s infamous 3-2 victory over Mississippi State:

GE = (1 / 7) / (31 / 2) = 0.143 / 15.5 = 0.0092

Granted it would have been the same if the score was 49-48, but still. Finally, here’s UNC’s 45-24 win over Boston College:

GE = (21 / 7) / (24 /2) = 3 / 12 = 0.2500

Fremeau then takes the game efficiency data and adjusts and expands his stats, factoring in strength of schedule as well as adjusted offensive and defensive efficiency based on opponents. Most or all of Fremeau’s statistical explanations can be found in his articles on Football Outsiders.

With this rubric in mind, Let’s look at his FEI top 25. Keep in mind, he factors out all games against FCS opponents.

Rank	Team	W-L	FEI
1	Texas	8-0	0.302
2	Penn State	8-0	0.284
3	North Carolina	5-2	0.277
4	USC	6-1	0.267
5	Florida	6-1	0.266
6	Alabama	8-0	0.245
7	Georgia	6-1	0.217
8	Virginia Tech	4-3	0.200
9	Georgia Tech	4-2	0.200
10	Oklahoma	6-1	0.200
11	Oklahoma State	6-1	0.197
12	Texas Tech	6-0	0.194
13	Florida State	4-1	0.192
14	Missouri	5-2	0.184
15	Ohio State	6-2	0.166
16	Connecticut	5-2	0.162
17	Mississippi	3-4	0.157
18	Vanderbilt	5-3	0.141
19	Boston College	4-2	0.138
20	Ball State	7-0	0.135
21	Miami	4-3	0.134
22	Pittsburgh	5-2	0.129
23	Iowa	4-3	0.128
24	Wake Forest	4-3	0.120
25	South Carolina	4-3	0.119

Wait a minute…am I reading this right? Is North Carolina…NUMBER 3!?!?

It’s like that, except replace “Michael Bay” with “the collective heads of everyone outside the ACC”.

Even more interesting than UNC’s #3 ranking here is their #1 ranking in adjusted defensive efficiency. What this means is that the Tar Heels do a better job of disrupting the efficiency of opposing offenses better than any other defense in the nation. And this is in spite of an apparent allergy to blitzes and the frustrating 2-minute defense. That, my friends, is impressive.

Mind you, FEI is not my new favorite poll because it’s very UNC-friendly. (But it helps.) It’s my new favorite poll because it is a lot less arbitrary than the coaches or media polls, and is a statistic-based ranking that better reveals who the good and bad teams really are.

Once again, this post could not have been possible without Tomahawk Nation.