FEI Week 7: Offensive and Defensive Efficiency Update

by Brian Fremeau

We have reached the midpoint of the 2015 season, the week in which all preseason projected data is removed from the FEI calculations and I begin to publish my full offensive, defensive, special teams, and field position ratings here at Football Outsiders. Opponent adjustments aren't based in part on five years of program power and recruiting success anymore. Every team is now evaluated on its 2015 efficiency results alone.

I have received a lot of feedback over the years that it is difficult to decipher the meaning of some of these stats I maintain and post each week. In particular, the raw and opponent-adjusted offensive and defensive efficiency metrics have always been presented in a way that was difficult for some readers to wrap their heads around. Oregon ranked No. 1 in offensive efficiency last year with a 0.872 OE rating, and No. 3 in opponent-adjusted offensive efficiency with a .747 OFEI rating. What do those numbers 0.872 and 0.747 even mean?

My idea was to present the data this way as a "percentage better (or worse) than average." Oregon, in other words, had raw game offensive performances that were 87.2 percent more efficient than an average team. When adjusted for the opposing defenses the Ducks faced, Oregon performed 74.7 percent better than an average team. Presenting the data this way made sense to me, but wasn't easily applied to actual results on the field.

An offense that is 75 percent better than average could be understood to score 75 percent more points over the course of a game than an average team, if we control for the strength of the opponent and starting field position. If an average team against an average opponent would score 24 points over the course of a 12 possession game, an offense that is 75 percent better than average should score 42 points over the course of a 12 possession game. Applying efficiency ratings at the game level isn't a particularly complicated set of mental gymnastics, but what does 75 percent better than average look like at the possession level?

I've been using game splits to break down the offensive, defensive, and special teams contributions to margin of victory or defeat. Even though the source of the calculations is complex, I think the game splits results are pretty tangible. A team wins by seven points, and game splits indicate that the team's offense generated 3.8 points of that seven-point margin, the defense generated 2.1 points, and special teams generated 1.1 points. Add those values up to get to the 7.0-point margin of victory.

Or take Alabama's 41-23 victory over Texas A&M this weekend. Where did the Crimson Tide's 18-point scoring margin come from? Not from the Alabama offense (minus-0.5 points generated), and not from the Alabama special teams (minus-15.2 points generated, due in large part to an A&M punt return touchdown, a blocked punt, and a fumbled Crimson Tide punt return). Alabama won the game on an enormous value generated by its defense -- 33.8 points primarily earned on three interception return touchdowns.

It occurred to me that since I represent offensive, defensive, and special teams game splits values in terms of contributions to the scoring margin of victory or defeat, that might be the best way to represent efficiency in those categories as well. Alabama's offense lost 0.04 points in scoring value per drive, its defense gained 2.11 points per drive, and its special teams lost 0.45 points per possession against the Aggies. The per possession averages for each team over the course of the season is the new representation of Offensive Efficiency (OE), Defensive Efficiency (DE), and Special Teams Efficiency (STE).

OE data is adjusted based on the strength of the opposing defenses faced to produce Opponent-adjusted Offensive Efficiency (OFEI). DE data is adjusted based on the strength of the opposing offenses faced to produce Opponent-Adjusted Defensive Efficiency (DFEI). I currently don't make opponent adjustments for special teams efficiency because the data is pretty noisy at the game level (one big play has a dramatic impact on game-level STE) and the opponent adjustments are unreliable as a result -- I may need to revisit this at a later date.

My goal is for the new offensive efficiency, defensive efficiency, and special teams efficiency ratings to be easier to digest and to make them more reliably scaled to one another. Positive values on offense, defense, and special teams all generate scoring value over average. Negative values are bad.

The No. 1 team in this week's FEI ratings is the Clemson Tigers, and their ratings are 0.53 OFEI (No. 23 national rank), 1.25 DFEI (No. 2), and 0.01 STE (No. 54). The ratings themselves mean that Clemson's offense generates 0.53 more points than average per drive, its defense generates 1.25 more points than average, and its special teams generates only 0.01 points per possession. I think the ratings clearly indicate that the per possession scoring values generated by the best offense (Baylor, 2.28 OFEI) and best defense (Alabama, 1.37 DFEI) are more significant than those generated by the best special teams (Michigan, 0.23 STE). Comparisons across offensive, defensive, and special teams units ought to make much more intuitive sense this way than in the old presentation of this data.

In addition to the offense, defense, and special teams data revised this week, I'm presenting field position data on the same per possession efficiency scale as well. To be more precise, I'm replacing what had been known as Field Position Advantage (FPA) with Field Value Efficiency (FVE). This too is based out of the game splits data, and represents not only the expected scoring value of each team's possessions based on starting field position, but also the value of defensive and special teams possessions that result in a turnover or score. As with special teams data, no opponent adjustments are made with FVE.

I have updated the 2015 pages for all of these stats, and have retained the supplementary metrics (available yards, value drives, field goal efficiency, etc.) that has been provided in years past as well. I have not yet published stat pages for previous years in the new format, in part because I'm hoping to get feedback from you on the format first. It should be noted that I expect the OE, OFEI, DE, and DFEI ratings and rankings of years past to change due to other revisions in the FEI formula I made this past summer.

Love the new numbers? Hate them? Let me know in the comments or drop me a message about the changes.

Degree of Difficulty Through Week 7

The DOD rankings are based on current FEI ratings, but instead of measuring efficiency against schedule, DOD measures record against schedule. How difficult would it be for an elite team (two standard deviations better than average) to play a given team's schedule to date and achieve that team's record?

My hypothesis is that the College Football Playoff selection committee is likely to value and reward something akin to DOD through their process and deliberations. Last year, the top four teams in DOD before the bowls were also the four teams selected for the playoff, and TCU and Baylor ranked fifth and sixth, respectively.

As of this week, the following teams have the most impressive records and have accomplished the most to date.

FEI Week 7 Ratings

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. Other definitions:

• SOS: Strength of Schedule, measured as the likelihood of an elite team going undefeated against the given team's regular season schedule. Schedule strength data based on FEI ratings and calculated across other dimensions can be explored in this interactive visualization.
• FBS MW: Mean Wins, the average number of games a team with the given FEI rating would be expected to win against its regular season schedule of FBS opponents.
• FBS RMW: Remaining Mean Wins, the average number of games a team with the given FEI rating would be expected to win against the remainder of its regular season schedule of FBS opponents.
• OFEI: Opponent-adjusted Offensive Efficiency value generated per possession.
• DFEI: Opponent-adjusted Defensive Efficiency value generated per opponent possession.
• STE: Special Teams Efficiency value generated per game possession.
• FVE: Field Value Efficiency value generated per game possession.

Preseason projection data receives no weight in this week's ratings. Ratings for all FBS teams can be found here.