Writers of Pro Football Prospectus 2008

Most Recent FO Features

WareDeM05-1.jpg

» Week 7 DVOA Ratings

Denver: great team, or the greatest team? Would you be satisfied with "one of the ten greatest teams?" Plus: hard times in the NFC South, where defense goes to die.

14 Oct 2009

Week 6 FEI Ratings

by Brian Fremeau

We're nearly halfway through the 2009 college football season, and what do we know? Better yet, how does what we know differ from what we thought we knew six weeks ago?

Only 13 of the current FEI top 25, and five of the top 10, were ranked as highly in the FEI projections at the start of the year. Florida has held onto the top ranking in FEI throughout the young season, but the Gators haven't yet performed at their stratospheric projected level and are clinging to a relatively narrow rating margin over No. 2 Alabama. No non-BCS teams appeared in the top 30 projected FEI; TCU and Boise State now reside among the top 10.

This is the final week in which Projected FEI data is factored into the FEI ratings, so it's a good time to examine the projections and recalibrate season expectations going forward.

By some measures, the projected data has performed pretty much as expected. Each week, the FEI ratings are used to produce Projected Win Expectations (PWE) for all future games. The weekly FEI Forecasts are designed to project the score at the conclusion of non-garbage possessions in a game, and PWE represents the likelihood of victory for the projected game winner. Table 1 provides a breakdown of the relationship between PWE and actual game outcomes so far.

PWE Range Win Loss Pct.
50% - 55% 14 11 .560
55% - 60% 15 9 .625
60% - 65% 9 12 .429
65% - 70% 13 9 .591
70% - 75% 16 8 .667
75% - 80% 27 8 .771
80% - 85% 18 5 .783
85% - 90% 18 9 .667
90% - 95% 24 2 .923
95% - 100% 57 4 .934

For some context, teams that are favored by 1-3 points according to Vegas (similar to 50-55 percent PWE) win about 54 percent of the time. This season, such teams are only winning at a 38 percent clip. Some sample sets of picks have performed well -- FEI is 31-17 (0.646) against the spread in games in which it projects the Vegas underdog to win outright; but overall, FEI Forecasts have been very pedestrian against the spread this year (.495). A quick scan of the performances of other computer rating systems suggests that the overall PWE straight-up record (.733) is off to a relatively good start. (Keeping pace with Bill Connelly's blazing S&P+ predictions is proving to be a much more lofty benchmark).

Based on PWE data for the entire the season, the FEI projections also provided a projected number of "Mean Wins" for every team -- the best approximation of each team's wins according to their own projected strength and that of their opponents. Though there has been plenty of turbulence in the FEI ratings over the first six weeks of 2009, roughly half of all teams have posted a record within 0.5 victories of their projected Mean Wins to date, and 75 percent have played to within one game. The biggest surprises are provided in Table 2.

Team FBS
Record
+/- Proj.
Mean Wins
To Date
New Proj.
Mean Wins
Delta Overall
Mean Wins
Idaho 5-1 +4.0 5.9 +4.1
Cincinnati 4-0 +2.3 8.7 +4.1
TCU 4-0 +1.8 9.3 +1.7
Miami 3-1 +1.6 8.0 +1.8
Houston 3-1 +1.6 8.6 +3.2
Florida State 1-4 -1.9 5.9 -0.2
Purdue 1-5 -1.9 4.8 -0.4
New Mexico 0-6 -2.2 2.5 -2.1
Vanderbilt 1-4 -2.6 3.4 -2.5
Ball State 0-5 -2.6 4.4 -1.4

Florida State is off to its worst ACC conference start ever (0-3) but is still hanging around just outside the FEI top 25. Its losses to No. 4 Miami, No. 12 Georgia Tech, and No. 14 South Florida were respectable (if gut-wrenching) and the Seminoles' blowout of BYU balances the loss to Boston College. Even though Bobby Bowden's team hasn't met its Mean Wins projection to date, the weakened ratings of several teams on the back half of their schedule has opened the door to salvage a modest record this year.

At the top of the table, Idaho has pulled itself out from the depths of despair (FEI projected the Vandals at No. 120 out of 120 teams), but the team still must face their toughest WAC opponents -- Nevada, Fresno State, and Boise State -- over the next month. Cincinnati travels to undefeated South Florida on Thursday night. If the Bearcats pass that test, Big East boredom may be their biggest obstacle to an undefeated campaign.

Week 6 FEI Top 25

The principles of the Fremeau Efficiency Index (FEI) can be found here. Like DVOA, FEI rewards playing well against good teams, win or lose, and punishes losing to poor teams more harshly than it rewards defeating poor teams. Unlike DVOA, it is drive-based, not play-by-play based, and it is specifically engineered to measure the college game.

FEI is the opponent-adjusted value of Game Efficiency (GE), a measurement of the success rate of a team scoring and preventing opponent scoring throughout the non-garbage-time possessions of a game. Like DVOA, it represents a team's efficiency value over average. Strength of Schedule (SOS) is calculated as the likelihood that an elite team would win every game on the given team's schedule.

Only games between FBS teams are considered in the FEI calculations. Since limited data is available at the beginning of the season, the ratings to date are a function of both actual games played and projected outcomes based on the 2009 Projected FEI Ratings. The weight given to projected outcomes will be reduced each week until mid-October, at which point the projections will be eliminated entirely. This is the final week in which projected data remains part of the calculations.

ALSO: FEI ratings for all 120 FBS teams are now listed in the stats page section of FootballOutsiders.com. Click here for current ratings, or here for 2008 ratings.

Rank Team FBS
W-L
FEI Last
Week
GE GE
Rank
SOS SOS
Rank
OE OE
Rank
DE DE
Rank
FPA FPA
Rank
1 Florida 4-0 0.272 1 0.356 4 0.473 103 0.321 25 -0.729 3 0.563 10
2 Alabama 6-0 0.255 6 0.358 3 0.246 42 0.401 18 -0.743 2 0.559 13
3 Virginia Tech 5-1 0.252 3 0.223 14 0.139 6 0.379 20 -0.456 18 0.539 24
4 Miami 3-1 0.218 4 0.003 57 0.123 4 0.146 39 0.100 65 0.511 48
5 Iowa 5-0 0.214 8 0.186 20 0.216 32 -0.056 59 -0.558 10 0.552 17
6 Texas 5-0 0.202 2 0.355 5 0.368 76 0.205 34 -0.688 4 0.564 8
7 TCU 4-0 0.190 5 0.192 17 0.383 83 -0.062 61 -0.526 11 0.547 20
8 USC 4-1 0.188 12 0.238 9 0.193 20 0.081 47 -0.759 1 0.524 34
9 Ohio State 5-1 0.186 11 0.236 10 0.278 51 0.073 49 -0.523 12 0.564 7
10 Boise State 4-0 0.182 9 0.402 1 0.596 114 0.452 16 -0.626 8 0.574 4
11 Cincinnati 4-0 0.179 15 0.343 6 0.278 52 1.312 1 -0.272 32 0.485 83
12 Georgia Tech 4-1 0.174 27 0.047 44 0.186 19 0.460 15 0.416 106 0.526 30
Rank Team FBS
W-L
FEI Last
Week
GE GE
Rank
SOS SOS
Rank
OE OE
Rank
DE DE
Rank
FPA FPA
Rank
13 Oklahoma 2-2 0.158 10 0.214 16 0.112 2 0.215 33 -0.499 14 0.527 29
14 South Florida 3-0 0.157 13 0.223 13 0.225 35 0.329 22 -0.568 9 0.500 60
15 Nebraska 4-1 0.156 24 0.383 2 0.230 37 0.525 10 -0.633 6 0.572 5
16 Kansas 4-0 0.154 16 0.281 7 0.305 57 0.703 6 -0.157 42 0.563 11
17 Auburn 5-1 0.150 7 0.144 24 0.292 54 0.499 13 -0.135 43 0.497 65
18 Pittsburgh 4-1 0.149 19 0.137 26 0.340 68 0.296 28 0.218 90 0.559 12
19 Oregon 5-1 0.141 21 0.215 15 0.212 26 0.079 48 -0.628 7 0.522 36
20 Arizona 2-2 0.141 18 -0.008 60 0.166 11 -0.104 67 0.117 69 0.519 41
21 Notre Dame 4-1 0.141 28 0.186 21 0.247 43 1.051 2 -0.027 54 0.448 103
22 Wisconsin 4-1 0.136 30 0.016 53 0.310 58 0.195 35 -0.032 52 0.479 88
23 Central Michigan 4-1 0.131 33 0.226 12 0.425 91 0.705 5 -0.346 26 0.500 63
24 LSU 5-1 0.131 20 0.124 28 0.171 12 -0.090 65 -0.391 22 0.529 28
25 West Virginia 3-1 0.130 14 0.151 22 0.243 41 0.393 19 -0.315 30 0.466 96

Posted by: Brian Fremeau on 14 Oct 2009

13 comments, Last at 15 Oct 2009, 11:36pm by Corey

Comments

1
by Bill Connelly :: Wed, 10/14/2009 - 11:43am

...I think I like your current FEI top 25 more than mine...I simply cannot shake Tennessee! They are to S&P+ this year what North Carolina was to FEI last year (and Philadelphia was to DVOA, I guess). Not only are they getting a major strength-of-schedule bump for having played Florida, but their trouncing of Georgia obviously made them jump up a bit more...they're in the top 5 now...

2
by DaninPhilly (not verified) :: Wed, 10/14/2009 - 11:52am

I think sos adjustments should be re-considered in general. It seems to me that most of the time when a computer ranking spits out a team with a much different ranking than is sensible, it boils down to a sos adjustment.

3
by J.D. (not verified) :: Wed, 10/14/2009 - 11:57am

Dumb question about the first sentence: wouldn't you expect about half the top 25 to be ranked higher than they were in the preseason, and half the top 25 to be lower?

10
by Kibbles :: Wed, 10/14/2009 - 11:56pm

If there were only 25 teams in college football, then yes. Since there are 120, then you would expect most of the teams currently in the top 25 to be ranked higher than they were in the preseason, because there's a lot more room for preseason top 25 teams to go down than there is to go up.

4
by peachy (not verified) :: Wed, 10/14/2009 - 12:13pm

Well, I like it. My Mumme poll ballot this past week was #1-11 & 13, so I feel all validated and stuff.

5
by Will :: Wed, 10/14/2009 - 12:55pm

Can't really argue with these rankings - they look good although Penn State is conspicuously missing. I guess that's what happens when you play only one better than terrible team and lose to them.

Will

6
by Scott P. (not verified) :: Wed, 10/14/2009 - 1:29pm

I'm still a bit hazy on how these rankings are calculated. Take Texas Tech, ranked 35th. They are 8th in the nation in game efficiency, 4th in Offensive Efficiency and 10th in Defensive Efficiency. Their SOS is 21st in the nation. How does that all add up to a 35th ranking? I am not saying they should be ranked higher, but am curious why they are not giving their other positions.

7
by mnaden :: Wed, 10/14/2009 - 2:09pm

As I understand these numbers, this would be my take on the Texas Tech question (as well as similar situations)......

The Efficiency numbers are all based on raw data while the Overall FEI is adjusted and based somewhat on the projections, so not only are the preseason projections of all the teams are still affecting the current overall rankings, but looking strictly at SOS ranking and trying to relate that to the Overall FEI is a little flawed

There are a couple of points to consider when looking at SOS. First, the SOS is for the entire season, so Texas Tech still has nearly all of their tough games still to go (Oklahoma, Oklahoma St., Nebraska, etc..), so they have not played the 21st ranked schedule to this point. Second, the appearance of teams such as Texas in their SOS is also a little deceiving because from what I understand, SOS is based on the probability of a top 5 caliber team going undefeated against that schedule. Texas Tech's performances so far have mainly been against weak competition (which DOES figure in to the drive calculations used in FEI), so once again, even though playing Texas does affect their SOS, it doesn't necessarily have the same impact on their FEI at this point.

As an example, if a team played cupcakes for the first 8 weeks of the season, but then ended with say Texas, Florida, and Alabama the final 3 weeks, they could easily be ranked at the top of all the raw efficiency categories through Week 8 but have a low FEI because of their competition (even though their SOS would be #1 with the teams remaining on their schedule).

Mainly, I think that the only time SOS will completely make sense in its relation to the FEI is at the end of the season after each teams' entire schedule has been played out, but each game that gets played from here on out will bring the two more in line with each other.

Hope this helps a little for now, and apologies if any of it is way off base, I'm sure Brian or someone else can offer a much better explanation.

8
by Todd S. :: Wed, 10/14/2009 - 2:19pm

Raise your hand if, before the season, you predicted Oklahoma and Texas back-to-back in Offensive Efficiency...at 33 and 34.

Also, since my alma mater is Ball State and I root for Purdue...bleah.

9
by Kal :: Wed, 10/14/2009 - 3:29pm

Assuming that USC and Oregon stay the course, that's going to be one heck of a fight. 1st ranked defense vs 7th ranked defense, 48th offense vs 47th offense. 36 and 34 FPA. Both teams have one game where their offense just didn't show up at all. And it's in Oregon...that's really going to come down to some crazy plays or injury luck.

11
by mrparker (not verified) :: Thu, 10/15/2009 - 6:19am

Brian,
I'm just curious. What would happen if you took out the top 10% drives adjusted for strength of opponent as well as the bottom 10% drives adjusted for strength of schedule. Wouldn't that help eliminate some of the bias from sos schedule adjustments?

12
by Sid :: Thu, 10/15/2009 - 6:02pm

What jumps out at me is Cincinnati having the best offense, and no other team being close.

13
by Corey (not verified) :: Thu, 10/15/2009 - 11:36pm

It really is confusing to post the SOS for the entire season. Why not post the current SOS? That would eliminate at least half the question regarding the rankings each week. What is the value in the season-long SOS anyway? You aren't posting season FEI projections anymore.