Writers of Pro Football Prospectus 2008

21 Sep 2016

FEI Week 3: Turnover Margins

by Brian Fremeau

The third weekend of the season featured several major statement victories from potential national championship contenders, including a handful of key road victories from teams at the top of the FEI ratings.

Alabama fell behind Ole Miss by 21 points in the first half in Oxford before rallying to a 48-43 victory, the largest comeback victory by the Crimson Tide under head coach Nick Saban and the program's first comeback win by 17 or more points since at least 2007. Michigan State dominated Notre Dame in South Bend for three quarters, mounting a 29-point lead before fending off a late rally to win 36-28. Ohio State rolled over Oklahoma in Norman. Houston danced with disaster at Cincinnati before dominating the fourth quarter to win by 24 points.

The road wins by the Spartans, Buckeyes, and Cougars -- along with another dial back in the weight given to preseason projections (now at 57 percent) -- pushed those teams into three of the top seven spots in this week's ratings, while Alabama remained on top by a healthy margin.

Four of the Power 5 conferences are represented in the FEI top 15. We have five teams from the SEC, four from the Big Ten, three from the ACC, and two from the Pac-12. The power conference absent from that tier is the Big 12, a league that has combined to win only two games to date against FEI top-50 opponents (Texas over Notre Dame in Week 1, and Oklahoma State over Pittsburgh in Week 3).

Texas fell on the road late Saturday night against California, a game in which they jumped out to 24-14 lead early in the second quarter, but then gave up touchdowns on four of the next five Bears possessions to fall behind before halftime. Two of those touchdowns came on California's best starting field position situations of the night -- a five-play, 46-yard score and a two-play, 24-yard score after a pair of Texas interceptions. There were no other turnovers in the game.

Turnover counts are a broad-brush statistic, but they're a popular and useful reference point for coaches and fans alike to explain the results of a given game. In FBS vs. FBS games since 2007, the team that recorded a positive turnover margin went on to win 74.5 percent of the time. That win rate jumps up to 82.5 percent when the turnover margin was plus-2 or better. (Turnovers are collectively defined here as fumbles that result in a change of possession, interceptions, and successful onside kick attempts).

For Texas, a two-turnover deficit has proven to be particularly costly. The Longhorns have had a turnover margin of minus-2 or worse in 26 FBS games since 2007, and they have won only two of those games (.077 win percentage). In all other games in that span, the Longhorns are 76-17 (.817). Among FBS teams in the last ten years, that's the biggest win percentage delta between minus-2 or worse turnover games and all other games.

FBS Record in Minus-2 Turnover Games and Other Games Since 2007
Team Minus-2 TO Pct Other Games Pct Delta
Texas 2-24 .077 76-17 .817 -.740
Florida 3-16 .158 75-19 .798 -.640
Middle Tennessee 2-28 .067 52-26 .667 -.600
Nebraska 7-26 .212 68-16 .810 -.597
USC 3-13 .188 82-23 .781 -.593

Another team worth mentioning in terms of poor turnover margins is Notre Dame. In their loss this weekend to Michigan State, the Irish coughed up the ball three times (a punt return miscue, a fumble, and an interception) and only forced one Spartans turnover. It was the 32nd occasion since 2007 that Notre Dame was minus-2 or worse in turnover margin. The Irish have managed to win a decent percentage of those games (.313) compared to the national average (.175), but they have had to dig themselves out of a deep turnover hole far too often. Among the top 60 teams in FBS win percentage since 2007, Notre Dame ranks behind only Nebraska (33 games) in the number of occasions in which it has been minus-2 or worse in turnover margin.

FEI Ratings Through Week 3

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.

FEI ratings through the first six weeks of the season are based in part on preseason projection data. Preseason ratings represent approximately 57 percent of this week's ratings. Ratings for all teams are linked here.

Rk Team Rec FEI GE Rk SOS Rk
1 Alabama 3-0 .329 .240 15 .109 12
2 Stanford 2-0 .275 .226 19 .211 41
3 Clemson 2-0 .259 .060 54 .328 60
4 Michigan State 1-0 .243 .085 43 .242 49
5 Ohio State 3-0 .233 .418 7 .139 16
6 LSU 1-1 .226 .006 63 .098 9
7 Houston 2-0 .222 .146 28 .408 77
8 Oregon 1-1 .201 .097 37 .188 35
9 Louisville 3-0 .195 .479 2 .158 21
10 Mississippi 0-2 .184 -.077 82 .056 5
11 Nebraska 3-0 .173 .239 16 .197 37
12 Arkansas 3-0 .170 .177 24 .101 10
13 Florida State 1-1 .168 -.204 104 .187 33
14 Georgia 2-0 .167 .053 56 .370 70
15 Michigan 3-0 .161 .385 9 .147 18
Rk Team Rec FEI GE Rk SOS Rk
16 Tennessee 3-0 .159 .137 32 .164 24
17 Texas A&M 2-0 .156 .094 39 .053 4
18 BYU 1-2 .156 -.008 68 .160 23
19 Oklahoma 1-2 .155 -.012 69 .202 40
20 UCLA 2-1 .147 .063 52 .187 34
21 Notre Dame 1-2 .141 .095 38 .215 43
22 Utah 2-0 .136 .140 30 .328 58
23 West Virginia 1-0 .122 .143 29 .372 71
24 Iowa 2-0 .121 .457 3 .460 84
25 Washington 2-0 .119 .557 1 .199 39
26 Wisconsin 3-0 .116 .170 25 .083 7
27 Baylor 2-0 .113 .286 13 .395 76
28 Georgia Tech 2-0 .110 .256 14 .259 52
29 Boise State 2-0 .110 .233 17 .694 118
30 Pittsburgh 1-1 .098 -.018 72 .222 44

Posted by: Brian Fremeau on 21 Sep 2016

8 comments, Last at 22 Sep 2016, 10:05pm by hoegher

Comments

1
by techvet :: Wed, 09/21/2016 - 4:07pm

"It was 32nd occasion since 2007 " should be "It was the 32nd occasion since 2007".

Regarding ND's poor turnover margin, going back to the Weis days, it be interesting to know how much is due to the lack of the defense getting turnovers.

2
by Brian Fremeau :: Wed, 09/21/2016 - 9:57pm

*thanks, edit fixed

4
by Brian Fremeau :: Wed, 09/21/2016 - 10:02pm

Irish have had 22 losses since 2007 in which they had a minus-2 or worse turnover margin in the game. On average, they generated 0.6 opponent turnovers per game in those 22 games and surrendered an average of 3.4 per game. In 11 of the 22 games, they did not force an opponent turnover. In their 10 wins since 2007 in which they had a minus-2 or worse turnover margin in the game, the Irish averaged 3.8 turnovers surrendered per game and forced 0.9 opponent turnovers per game.

3
by ramirez :: Wed, 09/21/2016 - 10:02pm

Brian,

Can you provide us with some insight into how you arrive at the rankings? I know last year's results are still a part of the formula. But how does Mississippi end up at 10th, when they're 5th in SOS but just 82nd in game efficiency?

5
by Brian Fremeau :: Wed, 09/21/2016 - 10:16pm

The SOS ratings are season-long schedule ratings, whereas the game efficiency ratings are results to date. And the FEI this week is 57% preseason projections and 43% results to date, So there isn't an easy GE+SOS=FEI thing happening at this time of year.

GE is game results to date, and if you pair that with the strength of the opponents played to date, you'll get a sense of the non-projection piece of the puzzle. One way to just eyeball those results is on my BCFToys FEI Ratings site where the records against opponent type are also provided.

Ole Miss has played No.1 and No.13, losing both games. So they don't have a very good GE as a result but they've played far tougher competition to date than anyone. 5th is their season-long SOS, but they'd be No.1 or No.2 easily in SOS to date. The only other team up there in SOS to date is Oklahoma (having played No.5 and No.7).

Basically then, the current FEI ratings are a combination of GE, opponent strength to date, and preseason projections. As I just typed that out I realized it might be helpful to actually post the ratings that way instead of confusing people with full-season SOS.

8
by hoegher :: Thu, 09/22/2016 - 10:05pm

You could have both, but maybe be more explicit on "full-season" vs "season-to-date."

6
by ammek :: Thu, 09/22/2016 - 3:38am

That wins-by-turnover-frequency chart in the link is full of interesting little quirks, thanks. It's amazing, for example, that Kansas has lost every game in which its turnover margin is +3.

I know you've done this before, but can you resume some of the causes of big differences between FEI and S&P+ early in the season? There are some real oddities: Michigan, Michigan State, Miami, Florida, Houston, Washington...

7
by Brian Fremeau :: Thu, 09/22/2016 - 12:15pm

Thanks for the suggestion, I'll try to tackle that next week.