Writers of Pro Football Prospectus 2008

Most Recent FO Features

WentzCar16-1.jpg

» Week 11 DVOA Ratings

DVOA has finally climbed on board the Wentz Wagon! The Eagles move into the No. 1 spot, but they aren't the only strong, well-balanced team in the NFL this year. New Orleans, Pittsburgh, and the Los Angeles Rams make this one of the best seasons ever for multiple teams over 30% in DVOA, and Minnesota isn't far behind.

16 Nov 2016

FEI Week 11: Upsets That Didn't Upset

by Brian Fremeau

A funny thing happened on the way to the postseason. One week ago, there were four undefeated teams from power conferences -- Alabama, Clemson, Michigan, and Washington -- gearing up for what figured to be a stretch run through November to an inevitable conclusion. None of those teams was a prohibitive favorite to win out, but each was reasonably capable of doing so based on efficient performances to date.

College football fans know that the only inevitability is chaos, of course, and Saturday's results only reinforced that axiom.

  • The Clemson Tigers, favored by 21.5 points according to betting lines and a 96.5 percent win likelihood according to FEI, were unable to stop Pittsburgh and fell to the Panthers on a late field goal by a final score of 43-42.
  • The Michigan Wolverines, favored by 24 points according to betting lines and a 92.6 percent win likelihood according to FEI, allowed the Iowa Hawkeyes to hang around and win on a late field goal by a final score of 14-13.
  • The Washington Huskies, favored by 10 points according to betting lines and an 87.5 percent win likelihood according to FEI, were overrun by the USC Trojans by a final score of 26-13.

If it wasn't the most stunning college football weekend in recent memory, it was close. In addition to those three results, six other teams that were underdogs by 10 or more points emerged victorious this weekend. In the entire 2015 season, including bowls, there were only 30 occasions all year in which an upset occurred in an FBS-versus-FBS game in which one team was favored by 10 or more points. Favorites by 10 or more points went 306-30 (91.1 percent) last year. To date this season, there have been 49 such upsets in FBS vs FBS games. Favorites by 10 or more points have gone 267-49 (84.5 percent) this year. Such favorites went 27-9 (75.0 percent) last week.

One of the things I look forward to most in the statistical work I do each week is plugging in chaotic data to see how dramatically the FEI ratings adjust as a result. In this case, however, not much changed. Alabama, Michigan, Ohio State, and Clemson were all ranked among the top four teams in last week's FEI ratings, and all four are still there this week (though the Wolverines and Buckeyes switched spots). Washington dropped all the way from No. 5 overall to No. 7 overall -- but barely; they're in a virtual dead heat with three other teams in the No. 5 through No. 8 positions.

Sure enough, the College Football Playoff selection committee found little reason to shake up their rankings this week as well. Clemson and Washington dropped two spots each in those standings and Michigan held steady. This weekend's upsets were dramatic in that they opened the door up for other contenders to potentially close the gap, but not so dramatic that they turned anyone's world upside down.

I haven't spent much time in this column yet this year commenting on the College Football Playoff selection process, but the committee's output this week was predictable in spots and very unpredictable in others. In their initial rankings releases, the committee had identified teams that generally matched the priorities of the ESPN Strength of Record (SOR) rankings and my FEI Degree of Difficulty (DOD) rankings. Both of these rankings are derivative of team efficiency ratings models, but they are designed to assess a team's record against the strength of its schedule rather than dominance or efficiency against that schedule.

I publish a weekly DOD top ten on my site through a few lenses. How difficult would it be for an elite team (two standard deviations better than average) to have the given team's record against that team's schedule to date? How difficult would it be for a very good team (1.5 standard deviations better than average) to do the same? How difficult would it be for a good team (one standard deviation better than average to do the same? The ESPN SOR model is designed similarly, but based on their Football Power Index.

I know that the playoff selection committee does not employ these kinds of ratings models directly in their process, and they claim that they don't even look at or collectively discuss ratings models such as these at all. I don't doubt that claim, but I do think they philosophically evaluate team resumes similarly to the way DOD and SOR are designed to evaluate teams. The committee rankings have tracked closely with DOD and SOR (for power conference teams in particular) in each of the first two seasons.

The committee was in a virtual lock step with DOD and SOR in its first few rankings releases this year. But they did deviate dramatically from what those models suggested this week for a couple of key teams. West Virginia (No. 6 in SOR and No. 9 in DOD) was ranked 14th by the selection committee yesterday -- more closely aligned with the Mountaineers team efficiency ratings (15th in FEI this week) than on the strength of its one-loss record. Michigan had slipped down to No. 6 in DOD and No. 7 in SOR this week, but the committee kept the Wolverines ranked in the top four -- again, more closely aligned with team efficiency ratings (No. 3 in FEI) than strength of record.

There is still a very limited amount of data we have at our disposal to attempt to predict precisely how the selection committee is going to handle its process going forward. And any attempt at making sense of the individual weekly rankings releases is not as critical as an attempt to make sense of the committee's final rankings released after the conference championship weekend concludes. Will the committee favor strength of record again this year, or will it deviate a bit from that trend line in the end as it has this week?

One other data point that I'm including in my DOD ratings page is the frequency of elite performances by each contender. I'm defining these as I did last season with my opponent-adjusted game efficiency GFEI ratings, in which every single game result for both the winner and loser is rated and ranked against all others. The very best opponent-adjusted results to date include Louisville's crushing defeat of Florida State back in Week 3, and Ohio State's waxing of Nebraska in Week 10. How often has each playoff contender recorded a GFEI rating in the 90th percentile or better?

Seven of the eight teams that made the playoff in 2014 and 2015 had recorded at least five 90th percentile GFEI ratings through conference championship week. The only team that reached the playoff without having done so was the 2014 Florida State Seminoles (only three 90th percentile victories). They were the only undefeated team that year heading into the postseason, and the committee ranked the Seminoles third overall in their final playoff release. The Ohio State Buckeyes that year edged out Baylor and TCU for the fourth playoff spot. The DOD rankings had that race as a virtual dead heat, but Ohio State had a clear edge in 90th percentile performances -- seven total compared with TCU's four and Baylor's two.

There are currently only four teams this season that have recorded at least five 90th percentile GFEI performances to date, and perhaps not coincidentally, those teams are all ranked in the top four according to the playoff selection committee this week. Alabama has nine, Michigan has six, and Ohio State and Clemson have five each. As with DOD, I'm not suggesting that the committee is directly or indirectly applying GFEI data in their process, but rather that they may be emphasizing elite single-game performances over other data.

Elite wins trump bad losses, or so it seems that is the case in the eyes of the playoff selection committee. In only a few more weeks, we'll know for sure.

FEI Ratings Through Week 11

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 10-0 .345 .318 4 .161 28 .73 19 1.79 1 .04 34 .210 30 .765 36 .9
2 Ohio State 9-1 .297 .339 2 .135 24 1.31 3 1.22 4 .04 36 .238 34 .566 9 1.7
3 Michigan 9-1 .271 .349 1 .115 19 1.24 4 1.00 5 .19 2 .378 57 .304 2 1.3
4 Clemson 8-1 .269 .153 17 .212 41 .63 23 .79 11 .03 44 .223 31 .950 88 2.0
5 Wisconsin 8-2 .207 .135 21 .080 9 .19 48 1.32 2 .01 62 .082 3 .971 103 1.9
6 Louisville 9-1 .205 .300 5 .227 46 1.16 5 .83 8 -.02 79 .282 43 .805 48 1.7
7 Washington 8-1 .205 .331 3 .381 77 1.12 7 .80 9 .09 15 .596 101 .640 16 1.4
8 Auburn 7-3 .203 .195 10 .052 2 .70 21 .72 13 .06 24 .257 38 .203 1 .1
9 Boise State 9-1 .190 .222 8 .511 100 .85 15 .79 10 -.09 115 .578 99 .883 63 1.8
10 Washington State 8-1 .187 .234 7 .247 49 1.01 10 .53 22 .03 45 .466 72 .531 7 1.0
11 LSU 5-3 .180 .128 23 .075 8 .34 32 1.27 3 -.12 121 .109 10 .686 26 1.3
12 Western Michigan 9-0 .179 .298 6 .672 121 1.51 1 .22 40 .01 58 .741 122 .907 72 1.8
13 Penn State 8-2 .179 .107 28 .112 17 .50 24 .42 27 .05 30 .114 11 .980 111 1.9
14 Colorado 7-2 .167 .131 22 .148 25 .33 34 .97 6 -.10 116 .204 28 .724 32 1.3
15 West Virginia 7-1 .163 .124 25 .421 84 .19 49 .60 18 .03 47 .494 75 .853 57 2.5
Rk Team Rec FEI GE Rk SOS Rk OFEI Rk DFEI Rk STE Rk SOP Rk SOR Rk MWR
16 Florida State 6-3 .158 .053 47 .125 21 .81 17 .03 59 -.04 86 .133 15 .938 82 1.7
17 Texas A&M 6-3 .156 .077 38 .062 3 .28 39 .13 48 .12 8 .078 2 .803 46 1.5
18 USC 7-3 .153 .155 16 .073 6 .44 26 .43 25 .12 9 .090 7 .807 49 1.4
19 Miami 5-4 .150 .098 29 .306 58 .02 65 .57 20 .07 22 .354 54 .865 59 1.5
20 Mississippi 4-5 .149 .036 52 .080 10 .97 13 -.12 71 -.04 88 .085 4 .942 84 1.7
21 Oklahoma 8-2 .139 .141 19 .181 33 1.07 9 .05 56 .00 69 .276 41 .658 19 1.1
22 Tennessee 6-3 .134 .048 50 .165 30 .42 29 .18 44 .08 19 .175 21 .942 83 1.7
23 BYU 5-4 .127 .049 48 .316 60 .08 58 .35 31 .12 7 .322 48 .980 114 1.9
24 North Carolina 6-3 .127 .056 46 .335 65 .41 30 -.36 91 .11 12 .352 52 .954 93 .8
25 Pittsburgh 5-4 .126 .001 64 .124 20 .83 16 -.49 100 -.02 80 .128 14 .969 101 1.8
26 Virginia Tech 6-3 .117 .094 30 .316 61 -.05 70 .42 28 .04 32 .381 61 .831 55 1.4
27 Troy 7-1 .117 .155 15 .329 63 .25 42 .61 17 .14 5 .340 50 .968 100 2.8
28 Northwestern 5-4 .117 .063 45 .127 22 .47 25 .37 29 .01 60 .137 16 .924 76 1.7
29 Utah 7-2 .113 .090 33 .315 59 .10 56 .30 34 .05 28 .457 71 .690 27 1.2
30 Houston 7-2 .112 .140 20 .422 85 .22 46 .56 21 .02 56 .639 109 .661 23 .8

Posted by: Brian Fremeau on 16 Nov 2016

7 comments, Last at 17 Nov 2016, 4:52pm by Eddo

Comments

1
by Subrata Sircar :: Wed, 11/16/2016 - 7:53pm

Michigan and OSU will shake themselves out in ten days - the loser will be out of the playoff hunt. In most cases, the winner will play Wisconsin and the B1G champion will end up in the playoff. (For example, a 11-1 Michigan team that beat OSU in Columbus, Wisconsin twice including a neutral site conference championship game, and PSU would have a comparable-or-better resume with any non-unbeaten team.)

The exception is if PSU and OSU both win out. Then we'll have a one-loss OSU squad which didn't even make the conference championship, and a PSU squad with wins over OSU and Wisconsin. A two-loss PSU could be ranked ahead of the one-loss OSU squad they beat, but barring catastrophe elsewhere, neither one should make the playoff over the other likely one-loss teams.

2
by Will :: Thu, 11/17/2016 - 2:56am

If Ohio State wins out and misses the B1G Title game, they will have wins over current CFP #3 Michigan, #7 @Wisconsin, #9 @Oklahoma, #18 Nebraska. Their loss would be #8 @Penn State.

If Penn State wins out, they will have wins over current CFP #2 Ohio State and #7 Wisconsin. Their losses would be #3 @Michigan and UR @Pitt.

One of those resumes is clearly better than the other.

Will

4
by Aaron Brooks Go... :: Thu, 11/17/2016 - 9:43am

There is precedent for a team which did not win its division making the top-2.

7
by Eddo :: Thu, 11/17/2016 - 4:52pm

In the committee era?

3
by annagrey :: Thu, 11/17/2016 - 7:46am

All the best to you.

5
by Raiderjoe :: Thu, 11/17/2016 - 1:04pm

thank you, you too..

6
by Raiderjoe :: Thu, 11/17/2016 - 1:06pm

if tema does not win own conference, them being in 4-team playoff seems every wrong and stupid to me,

8-team playofofs- fine. two big 12 or big 10 or pac 14 whatever. but not with 4 spots.

also, if w. Michigan goes undefeated, should be in playoffs. don't care about schedule. still fbs team.