Fremeau Efficiency Ratings

College football power ratings and analysis

FEI Week 8 Ratings and Adjusted Possession Advantage

by Brian Fremeau

FEI ratings are back this week after some work-life balance items got in the way last week. And we're not just returning to our regularly scheduled programming, we have much more data to dig into this week. Now that we have eight weeks of college football data to parse, I've eliminated preseason projected data from the FEI formula; introduced a new adjusted possession efficiency output; published offensive, defensive, and special teams ratings pages; and rolled out my opponent-adjusted single game performance ratings, GFEI.

We'll start with the new column in a prominent position on the FEI overall ratings page, Adjusted Possession Advantage (APA). Though I have made subtle changes to the FEI formula over the years, I've consistently kept its output in the same format, despite that format not having a clear translation to its meaning. An elite team may have an end-of-year FEI rating of 0.300, and average team would have a rating of 0.000, but what did those terms mean? I haven't changed this formatting (yet), in part because I haven't settled on an alternative. APA may be that alternative, but I'm going to run it alongside FEI this year before making an official decision on it.

APA represents the per-possession opponent-adjusted scoring advantage a team would have over an average opponent. An elite team with a .300 FEI rating might not be easily translated, but an elite team with a 2.97 APA rating (as Alabama has this week), can be more easily translated. On every possession exchange against an average opponent, Alabama would be expected to score nearly three more points than its opponent. That translational change alone has me strongly considering formatting FEI in this manner going forward. But APA isn't simply a conversion of FEI to a scoring-over-average value. It also represents a fundamental change in the way opponent-adjustments are made in my formula. In the simplest terms I can describe, FEI leans more heavily on single-game performances against top teams, and APA leans more heavily on season-long efficiency data rather than single game data.

Clemson ranks No. 1 overall in FEI this week, and No. 2 in APA. Alabama ranks No. 1 in APA and No. 2 in FEI. The reason for this flip is Clemson's most recent victory over previously undefeated North Carolina State, a 41-7 thrashing of the FEI No. 17 team in the nation. That game performance (.983 percentile) accounts for 23.4 percent of Clemson's FEI rating. Alabama hasn't had the opportunity to play a team of North Carolina State's caliber yet to date -- their 39-10 victory over FEI No. 23 Missouri is the toughest test they've faced -- and that victory only represents 17.0 percent of their overall FEI rating. FEI essentially gives Clemson extra credit for their dominance over the Wolfpack.

But APA doesn't view that game as significantly as does FEI. Alabama has been more consistently dominant than any other team this year, and APA finds that to be more impressive than a single dominant performance like Clemson's over North Carolina State. The Crimson Tide have single-game opponent-adjusted percentile performances through eight games as follows:

  • .597 vs FEI No. 119 Louisville
  • .794 vs FEI No. 105 Arkansas State
  • .990 vs FEI No. 61 Ole Miss
  • .966 vs FEI No. 31 Texas A&M
  • .903 vs FEI No. 129 Louisiana Lafayette
  • .975 vs FEI No. 73 Arkansas
  • .992 vs FEI No. 23 Missouri
  • .978 vs FEI No. 82 Tennessee

(Click here for GFEI single game performance data for every team.)

Alabama has had six single performances in the 90th percentile, whereas Clemson has had only three. APA considers Alabama's consistency in elite performances to be a better indicator of future success than Clemson's weighted single-game performance against a better team than Alabama has faced to date. I have a hunch that this change may be preferred by season's end as a better approach, but I'll likely need to produce and examine APA's retrodictive and predictive performances versus FEI in the off-season before committing to this change long-term. Again, for the remainder of this year, I'll run both columns each week.

As for my offensive, defensive, and special teams ratings pages, you'll notice a few more columns sprinkled into these as well. On the special teams page, I've dropped the starting field position columns that used to accompany these ratings since I now have that data split out now in a new format on my site. I've added in opponent field goal efficiency and team and opponent extra point efficiency in order to comprehensively represent each of the non-offensive and non-defensive values generated over the course of each game. I've also made SFEI an opponent-adjusted version of special teams efficiency, as I previously had not found a suitable way to represent those opponent adjustments.

The offensive and defensive pages will look more familiar, but I added a new column to each representing "Ball Control Rate." This was inspired by long-time reader and Oklahoma State fan David Hudson, who asked me recently to look into an issue he perceived to be plaguing his Cowboys. He suggested that their seemingly unusual propensity to be a boom-or-bust offense that, whether successful (explosive plays) or not (three-and-outs), put its defense in position to return to the field too quickly. I dug into a boom-or-bust offenses a while back, but I decided to flip the idea into the offenses that control possession, not in terms of time elapsed on the clock but rather in terms of the number of plays run before giving the ball back to its opponents. An elite offense like Alabama's, scoring touchdowns on 63.2 percent of non-garbage drives so far this year, is scoring very quickly. The Crimson Tide have the nation's best touchdown rate and the nation's best rate at avoiding three-and-outs, but they only rank 45th in ball control rate. North Carolina State and Army, both ranked among the top 10 in opponent-adjusted offense, rank No. 1 and No. 2 respectively in ball control rate.

Oklahoma State, meanwhile, has had more ball control success this year than Dave had suspected. The Cowboys rank 20th in ball control rate through Week 8, higher than they do in touchdown rate (26th) or its first down rate (29th). Perhaps a different measure of ball control would result in a different output, and perhaps I'll need to explore historic data in this category before drawing deeper conclusions. I'm open, as always, to feedback.

FEI Week 8 Ratings

The Fremeau Efficiency Index (FEI) is a college football rating system based on opponent-adjusted possession efficiency. Adjusted Possession Advantage (APA) ratings represent the per-possession scoring advantage a team would be expected to have on a neutral field against an average opponent, calculated as a function of current FEI overall, offense, defense, and special teams ratings.

Strength of Schedule ratings (PSOS) represent the average number of losses an elite team (two standard deviations better than average) would have against the team's regular season schedule to date. Offensive FEI (OFEI) is scoring value generated per drive adjusted for starting field position and opponent defenses faced. Defensive FEI (DFEI) is scoring value generated per opponent drive adjusted for starting field position and opponent offenses faced. Special Teams FEI (SFEI) is scoring value generated per possession by a team's non-offensive and non-defensive units adjusted for opponent special teams units faced. The team's record to date against opponents ranked in the FEI top 10 (v10), top 20 (v20), top 30 (v30), top 40 (v40), and top 50 (v50) are also provided.

Ratings and supporting data are calculated from the results of non-garbage possessions in FBS vs. FBS games.

Click here for ratings for all 130 FBS teams.

Rk Team Rec FEI APA Rk PSOS Rk OFEI Rk DFEI Rk SFEI Rk v10 v20 v30 v40 v50
1 Clemson 6-0 .271 2.33 2 .49 78 2.86 19 .8 1 .032 54 0-0 1-0 1-0 3-0 3-0
2 Alabama 8-0 .270 2.97 1 .34 92 4.21 1 .9 2 .062 40 0-0 0-0 1-0 2-0 2-0
3 LSU 6-1 .260 1.99 5 1.36 6 2.53 46 1.3 9 .344 1 1-1 3-1 4-1 4-1 4-1
4 Georgia 5-1 .248 2.30 3 1.04 13 3.62 3 1.6 21 .231 6 0-1 0-1 1-1 2-1 2-1
5 Oklahoma 6-1 .236 2.15 4 .59 60 4.14 2 2.2 68 .148 17 0-0 0-0 1-1 2-1 2-1
6 Michigan 7-1 .212 1.96 6 .94 20 2.81 21 1.1 3 .205 8 0-1 1-1 1-1 3-1 3-1
7 Washington 5-2 .209 1.72 7 .67 47 2.92 14 1.5 15 .014 57 0-0 0-0 1-1 1-2 2-2
8 Kentucky 5-1 .203 1.56 11 .83 29 2.32 56 1.3 8 .243 4 1-0 2-0 2-0 3-1 3-1
9 Florida 5-1 .196 1.57 10 .98 17 2.58 40 1.5 18 .243 3 1-1 2-1 2-1 2-1 2-1
10 Notre Dame 7-0 .195 1.60 8 .52 71 2.78 25 1.5 12 .040 48 1-0 1-0 1-0 2-0 3-0
11 Purdue 4-3 .183 1.42 17 .49 80 2.90 18 1.8 36 .050 43 0-0 1-0 1-1 1-2 1-3
12 Iowa 5-1 .182 1.59 9 .36 87 2.48 49 1.3 7 .239 5 0-0 0-0 1-0 1-1 1-1
13 Mississippi State 3-3 .166 1.54 12 1.38 5 2.57 41 1.2 5 .125 23 0-3 0-3 1-3 1-3 1-3
14 Duke 4-2 .165 1.06 29 .55 67 2.13 71 1.6 22 .040 49 0-0 0-0 0-1 2-1 2-2
15 Michigan State 4-3 .163 1.17 22 .86 28 2.14 70 1.5 13 .130 21 0-1 1-1 1-1 1-2 2-2
Rk Team Rec FEI APA Rk PSOS Rk OFEI Rk DFEI Rk SFEI Rk v10 v20 v30 v40 v50
16 Fresno State 5-1 .161 1.47 14 .13 124 2.90 16 1.4 11 -.030 74 0-0 0-0 0-0 0-0 0-0
17 N.C. State 4-1 .161 1.46 15 .81 33 3.21 9 1.9 40 .063 39 0-1 0-1 1-1 1-1 1-1
18 Ohio State 7-1 .160 1.53 13 .72 43 3.49 5 2.0 55 .099 33 0-0 1-1 1-1 1-1 1-1
19 Penn State 5-2 .151 1.35 18 .50 77 2.68 32 1.5 16 .109 30 0-0 0-2 0-2 0-2 1-2
20 Miami 4-2 .150 1.15 23 .77 37 2.18 67 1.2 6 -.059 78 0-1 0-1 0-2 0-2 0-2
21 Washington State 5-1 .149 1.24 19 .27 104 3.54 4 2.4 86 -.046 77 0-0 0-0 1-0 2-0 2-0
22 Iowa State 3-3 .145 1.18 20 .91 23 2.25 61 1.4 10 .123 25 0-1 0-2 1-2 1-2 1-2
23 Missouri 3-3 .145 1.43 16 1.49 3 2.92 15 1.6 26 .198 9 0-2 1-2 1-2 1-3 1-3
24 Auburn 4-3 .144 1.09 27 1.09 9 2.13 73 1.5 14 .165 12 1-1 1-2 1-2 1-2 1-2
25 Virginia 4-2 .143 .99 30 .86 27 2.18 68 1.6 27 .097 34 0-0 2-1 2-1 2-1 2-1
26 Texas 6-1 .137 1.08 28 .52 73 2.59 38 1.8 37 .133 19 1-0 1-0 1-0 1-0 1-0
27 West Virginia 4-1 .133 1.18 21 .50 76 2.85 20 1.8 35 .079 37 0-0 0-0 1-1 1-1 1-1
28 Utah 4-2 .129 1.11 25 .75 39 2.56 42 1.7 29 .148 16 0-1 0-1 0-2 1-2 1-2
29 Texas Tech 4-2 .127 .94 33 .34 95 2.69 30 2.0 56 .036 51 0-0 0-0 0-1 0-1 1-1
30 Cincinnati 5-1 .110 .82 37 .26 107 1.92 88 1.1 4 -.185 107 0-0 0-0 0-0 0-1 0-1

Comments

11 comments, Last at 26 Oct 2018, 11:25am

1 Re: FEI Week 8 Ratings and Adjusted Possession Advantage

by techvet // Oct 24, 2018 - 5:05pm

So I realize FCS games shouldn't be counted, but if, for example, Florida is playing two FCS teams this year, in what way is Florida "punished" for taking the easy way out? If Team X is 10-0 in FBS games and Team Y is 12-0 in FBS, shouldn't Team Y be rewarded for being "more daring" and playing teams at a higher risk level?

Login or Register to post comments

3 Re: FEI Week 8 Ratings and Adjusted Possession Advantage

by Brian Fremeau // Oct 25, 2018 - 7:47am

My calculations for schedule strength account for this, but the rating system isn't designed to "reward" any games against weak opponents, whether they are bottom-of-the-barrel FBS teams or FCS teams. I suppose Florida playing two FCS opponents allows them to not have the risk of a less-than-dominant performance against a lousy team, and that might contribute to an inflated rating. But the opposite could be true as well -- some teams may benefit from out-sized performances against weak opponents. I don't have confidence in FCS data (see here for my explanation: Notes) but this may be worth exploring further.

Login or Register to post comments

2 Re: FEI Week 8 Ratings and Adjusted Possession Advantage

by Chappy // Oct 24, 2018 - 5:13pm

Interesting. One thing I wonder about, that doesn't seem transparent to me, is how your system differs from S&P. I'll over generalize quite a bit, but it seems like, at least ranking wise, FEI is more bullish on the SEC and S&P is more bullish on the Big Ten. Why is this happening? I'll also note that one of the areas that is particularly stark is on defense. As a Michigan fan, I was trying to figure out why S&P has them #1 and you have them #3, but then I noticed that Alabama's D is #1 in FEI and #20 in S&P. This seems like a pretty big discrepancy. Is this due to something with adjusting for possessions? That is, Alabama's offense is so efficient that they leave the defense on the field alot and maybe FEI adjusts for this while S&P, which seems more total points based, does not? What am I missing?

Login or Register to post comments

4 Re: FEI Week 8 Ratings and Adjusted Possession Advantage

by Brian Fremeau // Oct 25, 2018 - 7:53am

Bill and I don't pry open each other's formulas, but I think discrepancies between S&P+ and FEI often come down to a few things:

1. We calculate garbage time differently. I'm not certain, but it is possible that Bill is measuring Alabama defensive performance in the second half of its games differently than I am. Anecdotally, the Crimson Tide have been exceptional until later in games, and I wonder if S&P is picking up on that drop off in a way that FEI is not.

2. Play success and drive success can be two very different things. A team may run four consecutive plays of 8-yard gains and then throw an interception. S&P might judge that to be four successful plays out of five, and FEI might treat that result similarly to a team that threw a pick on the second or third play of that sequence.

3. We have our own ways of calculating opponent adjustments. I get at that a little bit in today's column, that there are ways to emphasize single game performances versus season long performances with the same undelying data, but with different results.

Login or Register to post comments

10 Re: FEI Week 8 Ratings and Adjusted Possession Advantage

by Chappy // Oct 26, 2018 - 11:19am

Thanks. I think this all makes sense. I'll also note that, in S&P, Alabama is elite in success rate, but more like 20th in explosive plays. I certainly don't know how everything works, but I'd think that if a lot of those big plays come in what FEI considers garbage time, that would tell a relatively consistent story with above. With respect to your APA table above, I suspect a lot of the issue with SEC versus Big10 will become clearer as the season goes on. For example LSU has basically played all season against top 50 teams while Alabama has played 2. Seems like that match-up will uncover whether their team is historically great, or just a very good team propped up by an easy schedule.

Login or Register to post comments

6 Re: FEI Week 8 Ratings and Adjusted Possession Advantage

by Aaron Brooks Good Twin // Oct 25, 2018 - 10:23am

I wonder if it's because SEC teams almost always play at least one lower-division school and sometimes two (and this doesn't even include Tennessee). Whereas Big Ten schools only play Rutgers once.

Login or Register to post comments

11 Re: FEI Week 8 Ratings and Adjusted Possession Advantage

by Chappy // Oct 26, 2018 - 11:25am

I can sort of see that, but, at the same time, the table above shows the majority of top 30 SEC schools (notably excepting Alabama) have played a majority of games against top 50 teams. Most have winning records against top 50 teams too. In the Big Ten only Michigan and Michigan State have played most games against top 50 opponents--and only Michigan has a winning record against such opponents.

Login or Register to post comments

5 Re: FEI Week 8 Ratings and Adjusted Possession Advantage

by cstoos // Oct 25, 2018 - 8:48am

I'm curious how you distinguish between a good performance by team A and a bad performance by team B. Missouri is a great example.

Against Georgia, Missouri pretty much out played them the entire game, but just kept shooting themselves in the foot and gave Georgia the easy win. Against Alabama, early mistakes by Missouri probably made the point spread higher than it should have been, but Bama clearly would have won and totally outclassed Missouri, dominating the game.

So at face value those are two games, against the same opponent, with both teams winning by at least 2 scores. The realities were far different though.

Login or Register to post comments

7 Re: FEI Week 8 Ratings and Adjusted Possession Advantage

by Brian Fremeau // Oct 25, 2018 - 12:31pm

Comparing Georgia's performance against Missouri with Alabama's performance against Missouri, FEI sees Alabama's as better:

Georgia vs Missouri
.658 (percentile) GE
.963 GFEI

Alabama vs Missouri
.900 GE
.992 GFEI

Is that what you mean?

Login or Register to post comments

8 Re: FEI Week 8 Ratings and Adjusted Possession Advantage

by cstoos // Oct 25, 2018 - 2:39pm

Sorry. I meant how does APA account for things like that. Points gained, not from good performance by a team, but by bad performance by their opponent. Games like Georgia @ Missouri where Georgia struggled in general, but a few odd bounces made it look like they won by a blowout points-wise.

Login or Register to post comments

9 Re: FEI Week 8 Ratings and Adjusted Possession Advantage

by Brian Fremeau // Oct 25, 2018 - 8:22pm

One of the ways I account for the "bad bounces" is to not count turnover or special teams return yards in the ratings formula. Those returns certainly contribute to a team's margin of victory or defeat in a game, but they're not predictive or reliable indicators of team strength. That said, however, it seems you're trying to distinguish between the results of one team being non-reciprocal of the results of its opponent in the given game -- that is, an offense performing to a certain efficiency level in a given game being distinguished from the opponent defense performing to a certain efficiency level in the same game. I do not have a reliable way to account for that, no. In my case those results (after accounting for starting field position and turnover returns) are reciprocal efficiency rates. That's why when you review my Game Splits data, you'll see how the offensive and defensive values recorded for each game are reciprocally recorded for the opponent as well.

Login or Register to post comments