FEI Week 8: Possession Success Rate
by Brian Fremeau
Ohio State and Texas A&M fell from the ranks of the undefeated last weekend, but each still resides comfortably in the top 10 of the FEI ratings. That kind of outcome was something I used to need to defend as a feature of my drive efficiency ratings -- that a single game result from one week was not the most significant factor in the weekly ratings, but rather that new results produced each week simply help flesh out the overall profile of each team. That concept appears to now be embraced more widely in voted polls.
Ten years ago, the polls anchored teams in a more rigid ladder determined by the number in each team's loss column. Win and hold position, lose and fall behind the group of teams that didn't lose. After a 5-0 start in 2006, Auburn ranked No. 2 in the nation in the Associated Press poll before the Tigers suffered their first loss of the year to Arkansas. They dropped out of the top 10 the following week. Florida took their spot at No. 2 in the poll the next week, then lost to Auburn, a result that subsequently dropped the Gators seven spots behind each of the remaining power conference undefeated teams one week later.
In this week's AP top 25 poll, Ohio State dropped only four ranking spots following its loss to previously unranked Penn State, and the 6-1 Buckeyes still remain ranked ahead of three undefeated power conference teams (Nebraska, Baylor, and West Virginia). Texas A&M fell only three ranking positions after losing by 19 points to Alabama. This is not how polls used to behave, and I think it is an indicator that poll logic has shifted more towards the logic employed by our computer rating systems. Individual game results are meaningful, but cumulative season-to-date results mean so much more.
The polls no longer are a factor used to determine the championship game participants, of course, and it still remains to be seen if the College Football Playoff selection committee will consistently follow its own logic and ranking priorities year over year. The first release of the selection committee's top 25 rankings will come next week, and I anticipate that it will bear some resemblance to the AP poll and our efficiency ratings, at least at the top. Four undefeated teams and two one-loss teams -- Alabama, Michigan, Clemson, Washington, Ohio State, and Louisville -- are all ranked similarly this week by the Associated Press poll, as well as by Bill Connelly's S&P+ and my FEI rating systems.
Surprise game results this weekend could shake that group up a bit, but more likely than not those six teams will be identified next week by the committee as those best positioned for a playoff berth with a strong finish.
Like Ohio State, Louisville dropped only a few spots in the polls after suffering its first loss of the season to Clemson a few weeks ago. The Cardinals have been positioned as the top-ranked team with one loss ever since in the polls, and they rank right alongside the Buckeyes in the FEI ratings as well. Louisville's loss to Clemson ranks as the second-best loss of the year according to the opponent-adjusted single game efficiency ratings I produce, behind only a narrow loss by Troy on the road against Clemson in Week 2. Those are the only two losses ranked in the 90th percentile of GFEI results this year. For Louisville, it ranks as its third best single-game performance to date, behind its blowout wins over Florida State in Week 3 and North Carolina State last weekend.
Louisville's dominance, led by its Heisman-frontrunner quarterback Lamar Jackson, has been one of the stories of the season so far. Nobody wants to face the Cardinals, because they seem to be the kind of team that can't be stopped once they get rolling. In fact, they are better at that than any other team.
[ad placeholder 3]
I introduced a relatively simple stat last year I called possession wins, defined as a score when a team has the ball and a stop when it doesn't. I'm taking that concept a little bit further this year. Louisville has scored on 58.8 percent of its offensive non-garbage possessions this year, the third-best rate in the nation. The Cardinals have prevented a score on 80.5 percent of their opponent's offensive non-garbage possessions, the fourth-best rate nationally. Combined, Louisville has recorded a possession win on 121 out of the 175 total non-garbage possessions it has played this year -- a 69.1 percent overall possession success rate (third best nationally, behind Washington at 73.6 percent and Michigan at 71.8 percent).
Louisville has thrived most when stringing together consecutive possession wins. When Louisville's defense gets a stop, the Cardinals have scored on the ensuing possession 65.1 percent of the time, the best offensive "momentum possession success rate" in the nation. After an offensive score by Louisville, the defense has recorded a stop on the ensuing possession 86.7 percent of the time, the third best rate nationally. Combined, no team has a better overall momentum possession success rate -- stops are followed immediately by scores and scores are followed immediately by stops for Louisville on 73.6 percent of all momentum opportunities.
Louisville is far less prolific and efficient when it stumbles and records a possession loss. If the defense gives up a score, Louisville scores on the ensuing offensive possession only 26.7 percent of the time, the 82nd best rate nationally on "reset possessions." When the Cardinals fail to score on offense, their defensive possession success rate dips to 68.6 percent, the 32nd best rate nationally.
While Louisville is streaky, the Washington Huskies are exceptionally balanced in terms of possession success rates. Washington's offense ranks eighth nationally in momentum success rate (57.8 percent), first nationally in reset success rate (71.4 percent), and second in overall offensive possession success rate (61.0 percent). Their defense has top-10 ranks in all three categories as well. The Huskies have the nation's best overall possession success rate, and are ranked in the top three in both momentum possession opportunities and reset possession opportunities.
Possession success rate data for every team can be found here.
[ad placeholder 4]
FEI Ratings Through Week 8
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.
3 comments, Last at 27 Oct 2016, 11:57am
#2 by ramirez // Oct 26, 2016 - 9:00pm
I think I've figured out why I don't agree with some of the rankings produced by the FEI formula. It awards too big of a bonus for playing strong teams, regardless of the outcome of the game. To illustrate my point, let's use the example of the Mississippi Rebels. Mississippi is ranked 17th, and based on that, you would have no idea that they are a 3-4 team. Their 3 wins are not impressive. They beat an FCS team, a 2nd tier FBS team in Memphis, and a Georgia team that's no longer ranked in the ap top 25. If their 4 losses were actually impressive, I might buy the Quality Loss argument, but it doesn't work for Ole Miss.
The Rebels have lost to an Arkansas team that just got their doors blown off by Auburn, and just lost by 17 to LSU. But they have played a couple of good teams in Alabama and Florida State. The problem is that the losses to those teams weren't quality losses, either. Against Alabama, Ole Miss had a 24-3 lead late in the first half after getting a bunch of breaks and turnovers. Alabama then outscored them 45-6 to build an 18 point lead with 5 minutes left. Ole Miss then got a couple late TDs to cut the margin to 5, but that is not a quality loss, especially considering Mississippi was playing at home. And the other loss was to Florida State, a game which Mississippi lost by 11 on a neutral field, in which the Rebels were outgained by 200 yards and 11 first downs.
If you're going to rank Mississippi at 17 and FSU at 24, when FSU has a head-to-head win, there had better be a darn good reason. But the Seminoles' losses are no more damaging that those suffered by Ole Miss, and FSU's win over Miami is the most impressive for either team. It just doesn't pass the smell test. The FEI formula sees a close loss to Alabama, and another loss to FSU, and rewards Mississippi, despite the fact that they weren't impressive in either game. The other problem is that the FEI consistently overrates SEC teams, so Ole Miss gets another bonus because of their schedule, which isn't actually that difficult. Arkansas and Tennessee, for example, were hugely overrated by the FEI last year, which helped place a bunch of SEC teams at the top of the rankings to begin this season. It's time to revamp the formula, to more accurately reflect wins and losses.
Given the efforts be ESPN and CBS to shove the pro-SEC narrative down our throats every week, I"ll be very interested to see what happens at the end of the season. If Alabama remains undefeated and wins the SEC title, and Texas A&M breezes through the rest of their joke schedule to finish with one loss, we may see a strong push for the committee to put both of those teams in the playoff. If it comes down to a 1 loss A&M or a 1 version of a team like Clemson, Nebraska, or Washington, who do you think the committee will take?
#3 by Chappy // Oct 27, 2016 - 11:57am
Interesting article. However, I'm a little dubious about your claim that Louisville's offense and defense are synergistic in terms of momentum. The measure you are using doesn't, I think, control for field position. I think another alternative way to think of this is that maybe something that affects field position, like special teams is at play. I haven't watched Louisville much this year, but you have them ranked at 60, which isn't all that great relative to their peers. (A good kicker [and kick-off team] with a bad punter and set of return teams would seem to be consistent with your momentum story).