Writers of Pro Football Prospectus 2008

Most Recent FO Features


» 2017 Offensive Personnel Analysis

It's a three-receiver league, but for the first time since 2010, the frequency of 11 personnel actually went down last year. Was it a blip, or sign of things to come?

27 Sep 2017

FEI Week 4 Ratings

by Brian Fremeau

Like most that work with football analytics, my first interest in diving into numbers was to uncover information that the standard box score wasn't able to provide. I've discussed the origin story of the FEI ratings on several occasions, including as a podcast guest last week with Ed Feng's Football Analytics Show, and how the value of turnovers in particular drove me to collect data to better evaluate the results of a given game. A fumble or interception can be costly for an offense, but the variable costs of those turnovers can have a wildly variable impact on the results of a game.

Notre Dame defeated Michigan State on Saturday night by a final score of 38-18, a game in which the Irish jumped out to a 14-0 lead early on and were comfortably ahead for most of the remainder of the contest. That's one way to describe a 20-point victory. Another way is a glance at the team stats in the game. Michigan State outgained Notre Dame by 141 total yards (496 to 355) and earned five more first downs (26 to 21), but lost the turnover battle by three (two fumbles and an interception thrown by the Spartans, no turnovers by the Irish). Those numbers suggest that perhaps Michigan State would have won if not for the turnovers, a very different narrative altogether. Is the truth somewhere in between?

This game was a great example of several of the steps I take when breaking down possession data. My first step is to discard garbage time data altogether, and in this instance, my numbers don't count any of the drives recorded after 6:28 left in the third quarter. At that point in the game, Notre Dame held a 35-10 lead and had just forced a Michigan State punt.

By eliminating from consideration every drive after that point, I'm eliminating some of Michigan State's most "effective" drives in terms of yards accumulated -- 19 plays, 81 yards, and a turnover on downs; seven plays, 71 yards, and touchdown with around three minutes left in the game; nine plays, 82 yards as time expired. Over the whole game, the Spartans' offensive drives traversed nearly 500 yards, but nearly half of that came in garbage time. Notre Dame accumulated 73 percent of its yardage prior to garbage time. Even before we consider the impact of turnovers, the shift in perspective on the success each team had moving the ball in non-garbage time is significant, and highlights how misleading overall yardage and first down totals can be.

That said, the Irish were the beneficiary of three turnovers in their favor, and the value of two of them in particular was significant. I calculate the value of each turnover based on field position value, a combination of the value forfeited by the team with the ball and the value generated on the resulting field position of the team that took the ball away. On average across college football, each turnover is worth around four points in the exchange -- the offense coughs up half of that value in their lost scoring opportunity, and the resulting field position (often favorable, but not always) for the opponent accounts for the rest.

Against Michigan State, however, Notre Dame generated a total of 18.2 points in field position and scoring value on three turnovers, a 50 percent increase in value over what was expected. That's because two of the turnovers were exceptionally noteworthy in value lost and generated. The first was a pick-six on Michigan State's first offensive possession of the game. The Spartans were across midfield and faced a third-and-6 at the Irish 45-yard line. An incomplete pass would not have given Michigan State a scoring opportunity, but they still could have pinned Notre Dame deep with a punt, generating some potential value for Michigan State in the field position exchange. Instead, Irish cornerback Julian Love jumped the route and returned the ball for a touchdown, crediting the turnover with both the value of the score itself and eliminating the field position exchange value. In total, that turnover generated 7.8 points in scoring opportunity value by my game splits calculations.

The second turnover by Michigan State was more typical in terms of value generated. The Irish forced a fumble on another third-and-6 for Michigan State, this time in Michigan State territory. The offensive value lost by Michigan State on the drive-killing event was minimal, but the field position exchange setting up the Irish at the Spartans' 24-yard line was worth 3.9 points of scoring opportunity value.

The third turnover was another killer for Sparty. After Notre Dame capitalized to take a 21-7 lead, Michigan State marched downfield with a 74-yard drive, but the Irish forced a fumble at the goal line and jumped on the ball in the end zone for a touchback. In this case, the resulting field position exchange wasn't particularly beneficial, but the value of the turnover itself in terms of Michigan State's offensive value lost was huge. That play was worth 6.5 points in scoring opportunity value, and the Irish capitalized on it by driving the length of the field on the ensuing possession to extend the lead to 28-7 before the half.

The Irish had never previously capitalized on a turnover advantage in the Brian Kelly era (2010 to present) anywhere near as large as the one they had on Saturday night. That kind of turnover value success isn't likely to repeat itself often, of course, but defenses that generate some kind of net scoring value each game will find great success in the win column. Notre Dame is on a short list of teams that have generated net scoring value on defense in each of four games against FBS competition to date.

FEI 2017 Week 4 Ratings

The Fremeau Efficiency Index (FEI) is a college football rating system based on opponent-adjusted drive efficiency in FBS games. Preseason projected ratings are a function of five-year program ratings, recent recruiting success, and returning offensive and defensive experience, and account for 43 percent of this week's ratings. Strength of Schedule (SOS) ratings are a function of the projected FEI ratings of a given team's schedule of opponents and the location (home/away/neutral) of each game, representing the average number of losses an elite team (two standard deviations better than average) would have against the schedule.

Ratings for all 130 teams can be found here.

Rk Team Rec FEI SOS Rk
1 Alabama 4-0 0.279 1.06 52
2 Georgia 3-0 0.246 1.20 43
3 Clemson 4-0 0.244 1.14 47
4 TCU 3-0 0.228 1.25 39
5 Oklahoma 4-0 0.228 1.46 20
6 Florida State 0-2 0.212 1.76 2
7 USC 4-0 0.211 1.00 57
8 Ohio State 3-1 0.210 1.22 40
9 Michigan 4-0 0.197 1.30 34
10 Penn State 4-0 0.195 1.01 56
11 Notre Dame 3-1 0.179 1.47 18
12 Stanford 2-2 0.178 1.42 21
13 Washington 3-0 0.170 0.75 73
14 Oklahoma State 3-1 0.167 1.08 51
15 Wisconsin 3-0 0.160 0.62 82
Rk Team Rec FEI SOS Rk
16 Memphis 2-0 0.155 0.49 91
17 Auburn 2-1 0.145 1.72 3
18 Miami 1-0 0.135 0.98 58
19 Louisville 3-1 0.125 1.18 45
20 Oregon 2-1 0.124 1.04 53
21 Florida 2-1 0.118 1.39 27
22 San Diego State 3-0 0.115 0.39 107
23 Virginia Tech 3-0 0.109 0.90 64
24 UCLA 2-2 0.105 1.68 4
25 Houston 2-1 0.103 0.53 88
26 Mississippi State 2-1 0.098 1.54 10
27 Kansas State 1-1 0.092 1.26 36
28 LSU 2-1 0.090 1.39 28
29 Minnesota 3-0 0.077 0.83 69
30 Washington State 3-0 0.076 1.20 44

Posted by: Brian Fremeau on 27 Sep 2017