Writers of Pro Football Prospectus 2008

09 Nov 2016

FEI Week 10, FEI at 10

by Brian Fremeau

Today marks the 10-year anniversary of the first piece I published at Football Outsiders -- in fact, the first public introduction of the Fremeau Efficiency Index to the world, or at least this corner of it. A few months after Aaron Schatz and I first struck up an initial conversation about taking a deeper dive with college football data, he invited me to write that guest column to open up that dialogue to this community. I remain grateful for that opportunity and for the feedback loop it opened with the community here to take FEI from where it began to where it is today.

I have updated the opponent adjustment methodology a few times over the last ten years, but the core FEI input data has remained intact from the beginning. Game efficiency quantifies the success rate of a team scoring while in possession of the ball and preventing scores while not in possession of the ball over the competitive course of a game. I wrote that sentence ten years ago and I haven't changed that definition since. Game efficiency represents how a team played a given game, independent of the strength of its opposition.

The best possible game efficiency rating is one that would be recorded by a team that scores a touchdown on every offensive possession and stops its opponent from scoring on every opponent possession. The best game efficiency rating recorded this season nearly played out that way. Way back in the first week of the season, Michigan defeated Hawaii by a final score of 63-3. The Wolverines opened the game and their year by throwing an interception on their first play from scrimmage. But they scored a touchdown on each of the next nine drives and ran out the clock on their final drive of the game. In non-garbage time, Michigan led Hawaii 42-0 after 14 game possessions. The Wolverines scored a touchdown on six of their first seven drives, and didn't allow a point on any of Hawaii's first seven drives. Michigan's game efficiency rating of .857 in that contest is tied for the single best game efficiency rating of the season so far.

Michigan just crushed Maryland in Week 10 in similarly efficient fashion. The Wolverines scored a touchdown on each of their first five possessions -- in fact, on every one of their first-half possessions -- to take a 35-0 lead at halftime. Michigan tacked on a field goal on its first possession of the second half and held Maryland scoreless to that point in the game, a 38-0 lead in 13 non-garbage time possessions (six for Michigan, seven for Maryland). Michigan's game efficiency rating of .835 against the Terrapins ranks as the third-best game efficiency rating of the year to date.

Neither of those two opponents are any good, of course, and strong performances against weak teams can only tell us so much about the strength of a given team. I don't publish my opponent-adjustment algorithm. But I do show as much of the adjusted game results data that I can in order to help readers understand the priorities of the algorithm. The victory over Hawaii (No. 92 in this week's FEI ratings) has an opponent-adjusted single game efficiency ranking of 35th overall, and only accounts for 7.6 percent of Michigan's overall team rating.

It still ranks as an exceptionally efficient performance better than more than 95 percent of all single-game efficiency marks this year. But it isn't as meaningful for Michigan's overall rating as the Wolverine's performances against Colorado, Penn State, or Wisconsin -- the three teams ranked in the FEI top 20 that Michigan has faced and taken down this year. Each of those game results receives twice the weight of the Hawaii game result. As with the game efficiency fundamental data, the game weighting philosophy has remained intact over the last ten years -- FEI rewards teams for playing well against good teams, win or lose.

What has changed significantly over the last decade is the way in which I have carved up possession efficiency data to evaluate offenses, defenses, special teams, field position, and turnover values. The introductory 2006 piece not discuss unit efficiencies in any detail. In fact, I had barely begun to collect the data necessary to do anything of the sort at that time. Possession efficiency rates based on starting field position have since become fundamental components of the FEI companion statistics and my analysis of teams ever since the publication of that piece. And much of that came about due to suggestions and inquiries from readers of this column.

If you had told me back in 2006 that not only would FEI have matured over the next ten years in the manner that it has, but that I would be a regular contributor to Football Outsiders and ESPN Insider for most of that span, I probably wouldn't have believed you. If you had told me then that the network of college football stat analysts and advances in the field would expand so significantly over the next decade, and that I would get to collaborate with a guy like Bill Connelly and correspond with Sharon Katz and Ed Feng and so many others, I would have been ecstatic. Keep feeding suggestions and inquiries, and we'll keep developing them into better ways to evaluate the game.

FEI Ratings Through Week 10

The Fremeau Efficiency Index (FEI) is a college football rating system based on opponent-adjusted drive efficiency. Approximately 20,000 possessions are contested annually in FBS vs. FBS games. First-half clock-kills and end-of-game garbage drives and scores are filtered out. Unadjusted game efficiency (GE) is a measure of net success on non-garbage possessions, and opponent adjustments are calculated with special emphasis placed on quality performances against good teams, win or lose. Overall SOS ratings represent the likelihood than an elite team (two standard deviations better than average) would go undefeated against the given team's entire schedule.

Offensive FEI (OFEI) is value generated per offensive non-garbage possession adjusted for the strength of opponent defenses faced. Defensive FEI (DFEI) is value generated per opponent offensive non-garbage possession adjusted for the strength of opponent offenses faced. Special Teams Efficiency (STE) is the average value generated per non-garbage possession by a team's non-offensive and non-defensive units.

Strength of schedule ratings for games played to date (SOP) and for scheduled games remaining to be played (SOR), along with the projected number of FBS wins remaining against scheduled opponents (MWR) are also provided.

Ratings for all teams are linked here.

Rk Team Rec FEI GE Rk SOS Rk OFEI Rk DFEI Rk STE Rk SOP Rk SOR Rk MWR
1 Alabama 9-0 .348 .294 6 .159 29 .69 21 1.73 1 .03 43 .233 31 .680 38 1.8
2 Michigan 9-0 .300 .396 2 .131 23 1.61 1 1.04 6 .22 2 .479 67 .275 5 2.3
3 Ohio State 8-1 .295 .306 5 .128 22 1.25 4 1.11 4 .03 46 .273 38 .468 13 2.6
4 Clemson 8-0 .293 .174 12 .198 41 .64 22 1.03 7 .03 36 .222 27 .890 86 2.9
5 Washington 8-0 .237 .399 1 .431 82 1.45 3 .82 9 .11 11 .728 113 .592 22 2.4
6 Auburn 7-2 .235 .227 7 .044 2 .98 11 .75 11 .07 25 .231 29 .189 2 1.1
7 Louisville 8-1 .214 .311 3 .188 38 1.18 6 .77 10 .01 63 .241 34 .781 55 2.6
8 Wisconsin 7-2 .201 .104 30 .075 11 .12 55 1.28 3 -.01 74 .078 3 .953 111 2.9
9 Washington State 7-1 .183 .217 8 .245 48 .92 13 .51 20 .00 71 .520 73 .472 15 1.8
10 Texas A&M 6-2 .180 .087 33 .050 3 .39 30 .19 44 .10 13 .069 2 .729 44 2.3
11 Western Michigan 8-0 .179 .307 4 .697 123 1.55 2 .29 35 .00 66 .771 122 .903 93 2.7
12 Boise State 8-1 .178 .184 11 .494 94 .78 18 .69 13 -.10 112 .585 86 .843 74 2.7
13 Penn State 7-2 .174 .105 29 .103 17 .54 23 .45 25 .03 42 .111 7 .933 104 2.8
14 LSU 4-3 .171 .107 28 .057 5 .27 40 1.30 2 -.11 119 .103 5 .552 20 1.9
15 Colorado 6-2 .169 .121 25 .143 24 .31 35 1.09 5 -.14 125 .193 21 .740 46 2.3
Rk Team Rec FEI GE Rk SOS Rk OFEI Rk DFEI Rk STE Rk SOP Rk SOR Rk MWR
16 West Virginia 6-1 .151 .137 21 .443 85 .33 34 .48 23 .02 54 .570 83 .778 54 3.2
17 Virginia Tech 6-2 .149 .120 26 .406 75 .11 58 .36 30 .10 14 .462 64 .880 84 2.5
18 Florida State 5-3 .149 .005 61 .118 20 .78 19 -.08 65 -.04 93 .129 9 .915 97 2.7
19 Tennessee 5-3 .137 .019 55 .143 25 .31 36 .21 41 .10 17 .155 12 .926 101 2.7
20 Miami 4-4 .128 .079 36 .354 64 -.03 68 .51 21 .07 24 .415 54 .853 76 2.3
21 Mississippi 3-5 .127 .038 50 .067 8 1.01 9 -.15 76 -.02 76 .109 6 .617 26 1.8
22 BYU 5-4 .126 .049 45 .359 66 .04 64 .36 29 .12 9 .366 47 .980 126 2.0
23 Houston 6-2 .120 .144 20 .454 86 .26 42 .51 19 .02 51 .707 112 .642 28 1.8
24 North Carolina 6-2 .120 .063 42 .388 68 .36 32 -.17 80 .09 18 .431 58 .900 91 1.5
25 Oklahoma 7-2 .118 .121 24 .191 39 1.01 8 -.11 69 .01 62 .288 39 .662 33 1.8
26 USC 6-3 .111 .153 16 .068 10 .49 25 .15 45 .12 8 .165 16 .413 9 1.4
27 Northwestern 4-4 .103 .033 52 .144 26 .38 31 .34 32 .02 55 .160 15 .899 89 2.5
28 Troy 6-1 .100 .169 15 .276 52 .17 48 .52 18 .19 3 .294 41 .941 107 3.5
29 Toledo 6-2 .100 .188 10 .479 92 1.09 7 -.14 74 -.03 84 .732 116 .655 30 2.0
30 Oklahoma State 6-2 .093 .073 39 .414 77 .27 41 .05 56 .07 26 .629 97 .658 32 1.5

Posted by: Brian Fremeau on 09 Nov 2016

0 comments