Writers of Pro Football Prospectus 2008

Most Recent FO Features


» 2018 Free Agency Cost-Benefit Analysis

Is Kirk Cousins the best free-agent quarterback in recent memory? Should Trumaine Johnson or Malcolm Butler have gotten the larger contract? And what makes a free-agent contract good or bad, anyway?

21 Nov 2007

Fremeau Efficiency Index: Week 12

by Brian Fremeau

The Fremeau Efficiency Index (FEI) principles and methodology 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, 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.

Only games between FBS (Division 1-A) teams are considered. The Week 12 Ratings represent the results of games played through Saturday, November 17, 2007.

Rank Team W-L FEI Last Wk GE GE Rank
1 LSU (10-1) 0.264 1 0.259 6
2 Oregon (8-2) 0.258 2 0.204 10
3 West Virginia (9-1) 0.248 3 0.307 2
4 Ohio State (10-1) 0.246 4 0.291 3
5 South Florida (7-3) 0.223 10 0.170 16
6 Florida (7-3) 0.214 5 0.174 15
7 Arizona State (9-1) 0.210 8 0.202 11
8 Cincinnati (7-3) 0.205 6 0.155 17
9 USC (8-2) 0.201 12 0.201 12
10 Clemson (7-3) 0.200 7 0.215 9
11 Georgia (8-2) 0.198 11 0.097 28
12 Oklahoma (9-2) 0.188 9 0.274 5
Rank Team W-L FEI Last Wk GE GE Rank
13 Boston College (8-2) 0.186 17 0.132 21
14 Missouri (9-1) 0.185 13 0.217 8
15 Virginia Tech (8-2) 0.180 14 0.113 25
16 Illinois (8-3) 0.173 15 0.129 22
17 BYU (7-2) 0.162 21 0.188 13
18 California (6-5) 0.145 16 0.038 46
19 Kansas (10-0) 0.143 25 0.327 1
20 Auburn (6-4) 0.141 23 0.075 36
21 Connecticut (8-2) 0.140 18 0.135 20
22 Michigan (8-3) 0.140 26 0.089 31
23 Florida State (7-4) 0.139 22 0.066 38
24 Tennessee (8-3) 0.138 19 0.077 35
25 Virginia (9-2) 0.130 24 0.088 32

Offensive Efficiency (OE) represents the raw per-possession scoring rate for each team, and Defensive Efficiency (DE) represents that of its opponents. Adjusted Offensive Efficiency (AOE) and Adjusted Defensive Efficiency modifies OE and DE for field position and opponent. Field Position Advantage (FPA) represents the difference between a team's average offensive field position and its opponents'.

FEI Rank Team W-L OE OE Rank AOE AOE Rank DE DE Rank ADE ADE Rank FPA
1 LSU (10-1) 0.423 10 0.436 8 0.170 7 0.173 6 4.2
2 Oregon (8-2) 0.441 6 0.476 3 0.237 30 0.232 27 4.7
3 West Virginia (9-1) 0.440 7 0.435 9 0.201 17 0.196 15 5.0
4 Ohio State (10-1) 0.359 26 0.343 35 0.093 1 0.105 1 8.5
5 South Florida (7-3) 0.330 41 0.352 32 0.188 11 0.173 5 5.2
6 Florida (7-3) 0.521 1 0.559 1 0.344 91 0.319 73 4.4
7 Arizona State (9-1) 0.340 36 0.363 30 0.184 10 0.184 10 1.9
8 Cincinnati (7-3) 0.336 39 0.383 22 0.188 12 0.196 14 5.5
9 USC (8-2) 0.345 32 0.323 40 0.158 4 0.155 2 4.1
10 Clemson (7-3) 0.387 17 0.403 13 0.166 6 0.191 12 5.3
11 Georgia (8-2) 0.370 22 0.403 15 0.263 41 0.227 24 2.3
12 Oklahoma (9-2) 0.430 9 0.403 14 0.195 16 0.177 8 1.1
FEI Rank Team W-L OE OE Rank AOE AOE Rank DE DE Rank ADE ADE Rank FPA
13 Boston College (8-2) 0.314 46 0.394 18 0.183 9 0.222 19 -0.8
14 Missouri (9-1) 0.418 12 0.465 4 0.260 39 0.233 28 -0.7
15 Virginia Tech (8-2) 0.265 72 0.281 64 0.146 2 0.155 3 4.9
16 Illinois (8-3) 0.342 33 0.403 12 0.231 26 0.226 23 -2.0
17 BYU (7-2) 0.357 27 0.384 21 0.202 18 0.224 20 -0.6
18 California (6-5) 0.326 43 0.378 24 0.295 58 0.270 49 -1.2
19 Kansas (10-0) 0.447 5 0.371 26 0.156 3 0.166 4 4.3
20 Auburn (6-4) 0.287 61 0.318 42 0.221 25 0.183 9 -0.3
21 Connecticut (8-2) 0.276 67 0.289 61 0.174 8 0.192 13 1.6
22 Michigan (8-3) 0.292 57 0.300 53 0.192 14 0.176 7 2.6
23 Florida State (7-4) 0.236 90 0.260 76 0.202 19 0.218 18 5.0
24 Tennessee (8-3) 0.387 18 0.395 17 0.295 60 0.277 55 4.3
25 Virginia (9-2) 0.270 68 0.293 58 0.213 21 0.257 41 1.5

Click here for rankings of all 119 teams. Click here for Offensive/Defensive Efficiency rankings for all 119 teams.

In its Thursday night loss to Arizona, Oregon not only became the tenth Associated Press Top 5 team to lose to an unranked opponent this season, it joined South Florida, Ohio State, and Auburn as FEI top 5 teams to lose a game without falling in the FEI rankings. And like those others, the Ducks hold steady on the strength of their premier victories, two over FEI top 10 teams (Arizona State, USC). Unlike the others, the devastating injury to quarterback Dennis Dixon leaves Oregon in a much more precarious position heading into its final two games of the season. FEI doesn't make any adjustment for the Dixon injury, and his replacement, Brady Leaf, brings none of Dixon's Heisman-contending excitement to the table. But after falling behind by 20 points late in the first half following two Arizona touchdown returns, Leaf did lead enough scoring drives to get back in the game against an above-average Wildcat defense. It wasn't enough for a relief appearance victory, but with a bit more preparation, coach Mike Bellotti may be able to squeeze enough productivity out of the Dixon-less offense to beat UCLA and Oregon State. If not, FEI will drop the Ducks.

Arizona State and USC will face off Saturday in the biggest game of the weekend. In somewhat of a reversal of traditional Pac-10 play, the Sun Devils and Trojans are led this season not by high-flying offenses but by especially stout defenses. In last week's game between top 10 defenses, Ohio State topped Michigan 14-3. USC and ASU should produce a bit more offense, but unless the defenses create short fields off big turnovers, expect more slugfest than fireworks show.

In that other game of the weekend, the Kansas Jayhawks get to prove once and for all what they are capable of doing against above-average competition, facing No. 14 Missouri in Kansas City. Coming into this weekend, Kansas is one of only six teams that have yet to play an opponent int the FEI top 40 (the others are Hawaii, UTEP, SMU, Nevada, and Memphis). Fellow FEI top 25 teams, meanwhile, have played a combined 112 games against FEI top 40 opposition, winning 68 of those games. Against their toughest defense faced to date (Kansas State), the Jayhawks scored only twice in long-field situations. Against their toughest offense faced to date (Oklahoma State), the Jayhawks gave up four long-field touchdowns. Missouri boasts a better offense and defense than either of those units. A BCS championship game berth may be in its sights, but Kansas hasn't seen any team yet this season anywhere near as good as Missouri, and may need to make adjustments right out of the gate.


Boise State (9-1; No. 19 BCS, No. 17 AP, No. 42 FEI)
Even with a win over Hawaii on Friday night, the Broncos probably won't stake a claim to one of the coveted BCS slots this season. That's probably a good thing for Boise State, though, since in their only matchup to date with BCS competition, and their only game against a team rated by FEI in the top half of college football, the Broncos lost by two touchdowns to 4-7 Washington. Compare that to last season, when long before their Fiesta Bowl showcase victory, BSU dominated 9-4 Oregon State and 7-5 Utah out of conference. This year's Boise State schedule includes blowout victories over four of the worst six teams in the country. They should still beat Hawaii (now leading all teams in wins over Pythagorean Win Percentage, +1.97 wins), but even in a year in which every team is flawed, Boise State has done little to merit a bid for a big-time bowl.

Home Field Advantage

In the 2003 to 2006 seasons, home teams won 60.4 percent of games played between FBS opponents, and had an average margin of victory of nearly six points per game. Why, then, does FEI not factor in home field advantage in its formula?

The first consideration is whether such home field advantage measurements are biased. Unlike the NFL, college teams determine their own non-conference schedules, and more often than not, the power conference teams schedule non-power conference games on their own turf. Since those games do not represent an equal balance of home and away contests (the Big Ten is not playing MAC teams on the road), those home-dominated results skew the overall averages. Disregarding all non-conference games in the four-year span, the home team winning percentage dips to 56.9 percent, with an average MOV of 3.9 points per game.

A second consideration is whether home field advantage is meaningful in close games. From 2003 to 2006, in games decided by seven points or fewer, home teams had a record of 408-419 (49.3 percent). In conference games over that span, home teams had a record of 267-276 (49.2 percent) in such games.

Is home field advantage a factor in large-margin games? There certainly appears to be evidence of this in garbage time scoring. In the 1,556 home team victories from 2003 to 2006, home teams outscored the road teams in garbage time alone by 1,649 points. In these games, home teams outscored their opponents in garbage time in 385 games and were outscored in garbage time in only 188 games. In road team victories, home teams outscored their opponents in garbage time in 92 games and were outscored by the road team in 149 games. Thegarbage time scoring advantage for home victors was more than twice that of road victors over that span.

Since FEI already discounts garbage time scoring and possessions entirely, large margin game home field advantage is negated. Until more evidence can be collected to determine the effect, if any, of home field advantage in close games, FEI will continue to not adjust close game results.

Posted by: Brian Fremeau on 21 Nov 2007

12 comments, Last at 28 Nov 2007, 10:11am by JKL


by NewsToTom (not verified) :: Wed, 11/21/2007 - 4:22pm

My biggest problem is I still think the strength of schedule adjustments are out of whack. Here's a specific team issue:

Last week, USC's GE was .201 and its FEI was .190. USC didn't play last week, so their GE is still .201, but their FEI rose to .201. This strikes me as a relatively big change, particularly this late in the season and thus have most of the information we're going to get. What specifically happened this weekend that tells us USC is a better team than we thought they were a week ago?

by Pat (not verified) :: Wed, 11/21/2007 - 4:44pm

They should still beat Hawaii (now leading all teams in wins over Pythagorean Win Percentage, +1.97 wins)

What's funny is this statement comes right before a treatise on why home field advantage isn't used. Hawaii has the largest home field advantage in all of college football.

A second consideration is whether home field advantage is meaningful in close games.

This is, in all likelihood, a meaningless statement: games decided by 7 points or less are random - they're +/- 1 score, and any game that can be decided by a single play can't be definitive. It's not that HFA doesn't exist in those games. Its that its effect can't be resolved since it's swamped by the imprecision of the game itself.

If it's there for large-margin games, it's in all likelihood there for small-margin games, just not resolvable.

by AVTom (not verified) :: Wed, 11/21/2007 - 4:50pm

Great stats. Minor correction, the 'SC game is on Thursday.

Also, I dispute your contention that a strong defense is a "reversal of traditional Pac-10 play". For at least the last six or seven years, hasn't 'SC put up really good defensive teams? I guess taking the conference as a whole and looking at it historically, you could say that, but I've found the idea that the Pac-10 doesn't play defense to be a bit of an old-wives tale.

by Eric J (not verified) :: Wed, 11/21/2007 - 4:50pm

Just noticed that Navy has the #2 offense by this metric. Of course, they're still only ranked #70, on the strength of their #119 defense.

by NewsToTom (not verified) :: Wed, 11/21/2007 - 6:32pm

One commenter last week suggested that the strength of schedule adjustment looked about three times too large. With that in mind, I ran the FEI Top 25, less 2/3 of the difference between FEI and GE:

1. 3 West Virginia 0.2873
2. 4 Ohio State 0.2760
3. 19 Kansas 0.2657
4. 1 LSU 0.2607
5. 12 Oklahoma 0.2453
6. 2 Oregon 0.2220
7. 10 Clemson 0.2100
8. 14 Missouri 0.2063
9. 7 Arizona State 0.2047
10. 9 USC 0.2010
11. 5 South Florida 0.1877
12. 6 Florida 0.1873
13. 17 BYU 0.1793
14. 8 Cincinnati 0.1717
15. 13 Boston College 0.1500
16. 16 Illinois 0.1437
17. 21 Connecticut 0.1367
18. 15 Virginia Tech 0.1353
19. 11 Georgia 0.1307
20. 22 Michigan 0.1060
21. 25 Virginia 0.1020
22. 24 Tennessee 0.0973
23. 20 Auburn 0.0970
24. 23 Florida State 0.0903
25. 18 California 0.0737

Looking outside the FEI Top 25, Hawaii slots between Cincinnati and BC, while Boise State ends up between Oregon and Clemson. I'd run the full list of I-A teams and post that top 25, but can't parse the data on Brian's website successfully.

Subjectively, I'm not sure whether or not I like this any better than the current FEI. Boise hasn't looked to me like the #7 team in the country, BYU seems really high (admittedly, I haven't seen the Cougars play much), Hawaii's probably a little high, and Georgia looks too low at #19.

One possibility could be to use general conference-related adjustments, such that game against BCS conference teams are weighted twice as much for SOS purposes as non-BCS teams. That's a little crude, I know, but it does strike me as capturing a little better the institutional difference present in I-A.

by David (not verified) :: Wed, 11/21/2007 - 7:15pm

dude no offense but with the less strength of schedule adjustment your rankings look much much better

by Brian (not verified) :: Wed, 11/21/2007 - 7:48pm

There is a severe bias in your thinking on home field advantage in "close" games.

Because of home field advantage, when road teams do win, the scores will tend to be closer than otherwise. This gives the illusion of a weak home field advantage in "close" games, when in reality, what you're seeing is a real and significant home field advantage at work. The illusion is created by selecting close games to begin with.

I buy your argument about power teams scheduling weak out-of-conference teams at home. But the "meaningful in close games" argument is backwards thinking. You have to consider that home field advantage makes road victories "close."

by pm (not verified) :: Wed, 11/21/2007 - 8:17pm

This system continually shows that it is a joke. South Florida isn't a top 5 team. They shouldn't even be close to teh top 20. THe system can't be taken serious if keeps spouting off garbage rankings that are lot worse than even the worst of pundits.

by Dennis (not verified) :: Wed, 11/21/2007 - 8:20pm

I agree with those who think strength of schedule is weighted too much. It seems like teams can move too much when they don't play.

by DBM (not verified) :: Fri, 11/23/2007 - 7:49pm

I disagree with arguments for strenght of schedule being weighted too much.

David argues that the rankings with less weight on strength of schedule “look much much better.� That is not much of an argument. Does he mean that they look more like the BCS rankings? If so, then the BCS rankings may well be poor. Does he mean that they look more like the teams’ records (e.g., undefeated team Kansas moves up)? If so, I would argue that teams with better records may well have better records because they played weaker teams. Or does he have some other way of determining which rankings “look� better? If so, perhaps he can let us know how he makes these determinations, since they may well be valid. Given his argument as stated, however, it is impossible to lend it credence.

Dennis argues that a team’s position moves “too much� when the team doesn't play. This does not, however, mean to me that this makes the ranking system bad. After all, we have more information about the team’s opponents which allows us to make a better evaluation of the team’s prior performance. If a team (say, Appalachian State) beats a highly-ranked team (say, Michigan), then we might rank the first team quite highly as a result. However, if the highly-ranked team (Michigan) gets drubbed the following week (say, 39–7), maybe the first team’s performance is no longer quite so impressive after all, since the highly-ranked team probably shouldn’t have been so highly ranked. I would argue that the real measure of the quality the FEI (or any ranking method) is whether it serves as a good predictor of future performance.

by Cael (not verified) :: Sun, 11/25/2007 - 1:06am

Concur with #7:

Imagine home field advantage is worth 3 points. Assume that an average team at a neutral field would win about as many close games as it lost.

Now play those games at home; the games that at the neutral field were won by 5-7 points are now won by 8-10 and are no longer counted as "close." Games lost by 0-3 points are now won, and games lost by 8-10, and therefore not close are now only lost by 5-7 and are close.

There's no good reason to expect that the number of games that moved out of the close+won category is much different than the number that move in, and no reason to expect the number of games moving out of close+lost to be much different than the number that move in. If the average team wins as many close games as it loses at a neutral field, we should expect the same at home, no matter how large the home field advantage is.

by JKL (not verified) :: Wed, 11/28/2007 - 10:11am

I agree with Pat in #2, Brian in #7, and Cael in #11 regarding the home field advantage.

The key is your 56.9% winning percentage in conference games with average MOV of +3.9. You make it sound like 56.9% is insignificant, when it is not. Also, conference games still include a fair amount of mismatches where the road team will be a heavy favorite, so I would expect the home team winning percentage in games between relatively similar opponents to be even higher.

"Unlike the NFL, college teams determine their own non-conference schedules, and more often than not, the power conference teams schedule non-power conference games on their own turf. Since those games do not represent an equal balance of home and away contests . . ."

And this is why home field advantage should be accounted for, and not a reason to not include it. Some BCS teams scheduled four home games, some played two road or neutral games. As I pointed out in a previous week, the SEC non-conf schedule this year was the most skewed in this regard, where they played good, or at least other BCS teams, at home, but very rarely on the road. The two road contests were Tennessee at Cal and Miss State at West Virginia. If you don't account for it, you will overvalue whatever conference has a higher distribution of home games against quality opponents, and it may not always be the SEC.