Fremeau Efficiency Ratings
College football power ratings and analysis

FEI 2020 Projections

Trevor Lawrence
Photo: USA Today Sports Images

Let’s roll back to January. The LSU Tigers are making confetti angels on the field of the Superdome, basking in the glory of a dominant stretch run to close out the 2019 season, one of the greatest and most accomplished in college football history. The runner-up Clemson Tigers are immediately named national championship favorites in 2020, loaded with an abundance of returning talent and a mission to close out unfinished business. Alabama, Georgia, and Ohio State are filling in elite recruiting classes to maintain their respective positions as perennial contenders themselves. College football settles into its offseason rhythms, and fan bases for Power 5 and Group of 5 conferences alike are imagining the possibilities ahead.

The threat of a global pandemic hadn’t yet disrupted anything in our college football universe back in January. The health crisis itself has claimed hundreds of thousands of lives and has impacted so many more. College football as we knew it is a much more trivial casualty, of course. And yet, can you imagine going back in time and explaining to your January self the series of events and decisions that would lead to the cancellation of hundreds of non-conference games, the SEC moving to a 10-game conference schedule, and Notre Dame painting an ACC logo on its football field? I cannot.

If I’m being perfectly honest, I haven’t had the time or energy to work through the permutations of the offseason and their impact on my numbers. In July, just as we were putting the finishing touches on the Football Outsiders Almanac, the Big Ten announced that it would play an exclusive, conference-only schedule. We didn’t stop the presses and make rapid adjustments to our algorithms in that moment. Instead, we put our projections out as if the season were going to proceed as normal. Two months later, I’ve lost track of the number of times news has broken that might prompt another numbers adjustment. Multiple conferences have canceled games altogether, multiple players have opted out for the season, and multiple teams have postponed games in the coming days due to coronavirus.

Even if we could hit pause on the breaking news and confidently say the games scheduled from this point forward will happen, I still have significant concerns about what to do with the FEI ratings. A college football ratings system designed to rate the relative possession efficiency of 130 teams requires a baseline that may not be reliable in 2020 – what is an average team in that environment, and is it the same as an average team in any other year? A reliable rating system requires significant connectivity between teams to make sense of the relative strengths of programs that only play a handful of non-conference opponents. What happens when they play none? Is there any way to make sense of the SEC championship race in the context of the national championship race if they only play games against one another?

I haven’t yet answered these questions, and I may be working through possible solutions (adequate or otherwise) as the season rolls forward. Such is life in 2020.

Let’s roll back to January and start over. I decided to stop chasing my tail with the latest changes to the season schedule and [gestures at everything] and decided to make things simpler. My starting point each offseason as I build the FEI projections is weighted, five-year performance data. How did each team perform in terms of opponent-adjusted possession efficiency over the last five seasons? Why five? It works well as a baseline, and has since I first starting working with projection data over a decade ago. And in terms of correlation with next year success as measured by team efficiency ratings and individual game outcomes, it’s better than every other baseline I’ve tested.

Presented below are the 2020 FEI Preseason Projections, based entirely on information known as of the conclusion of last year’s championship game. I’ve kept all FBS teams in the ratings, partly because there’s still a chance some of the teams that aren’t planning to play this fall may yet still pull together a plan to do so, and partly because I want a record of the relative strength of all teams heading into this season. I haven’t decided whether I’ll present the full set of 130 FBS teams going forward (and have some teams hold their position in perpetuity), and I’m considering conference-specific FEI outputs to play with this year as well.

I did update projected mean wins and schedule strength calculations based only on the 77 teams with scheduled games this fall – games scheduled, that is, as of about a week ago when I last updated the spreadsheet. Yes, there have since been postponements, sigh. There are only 382 FBS games on the schedule this year, about half as many as we get in a full season, and every one of them is precious.

Schedule strength will be an interesting conversation this year, one that I think my system and method for calculating is well-suited to address. Teams are playing wildly different numbers of games – Air Force is scheduled to play only twice; two teams, Texas State and UL-Monroe have 12 FBS opponents each. My strength of schedule output measures the average number of losses an elite team (two standard deviations better than average) would be expected to have against the entire schedule. I’m guessing this schedule strength measure will be among the most intriguing to watch change over the course of the year, as every team’s schedule at this point can be confidently described as merely tentative.

I’ve also projected the likelihood of each team winning its conference, which assumes a conference schedule played in full, and the likelihood of each team earning a berth in the College Football Playoff field. The latter is based off the formula we use in the Football Outsiders Almanac projections. It remains to be seen if the CFP committee uses similar logic to set the field as they have in the past, or if they will be wildly unpredictable like everything else this year.

I hope I’m not undercutting my own credibility by putting this version of my 2020 numbers out there in this manner, but reverting back to performance history and blocking out all the other noise makes the most sense to me as I approach the year ahead. It feels like the only baseline I can count on. Since many teams aren’t playing right out of the gate, I’ll hold off on posting new numbers and an FEI Ratings column again for a few more weeks until we have a solid number of game results under our belt. In the meantime, here’s hoping each of you, your families and friends, and the sport we love together stays healthy and safe.

2020 FEI Projections

FEI Ratings projections are based on weighted five-year FEI ratings. FEI ratings (FEI) represent the per-possession scoring advantage a team would be expected to have on a neutral field against an average opponent. OFEI Offense ratings (OFEI) and DFEI Defense ratings (DFEI) represent the per-possession scoring advantages projected for each unit. Remaining columns for each team detail the number of FBS opponents scheduled (G), Mean Wins projection against scheduled FBS opponents, strength of schedule (SOS) calculated as the average number of losses a team two standard deviations better than average would be expected to have against the schedule, the percent likelihood of winning a conference championship (Conf), and the percent likelihood of earning a College Football Playoff berth (CFP).

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

Rk Team FEI OFEI Rk DFEI Rk G Mean
Wins
SOS Rk Conf CFP
1 Alabama 1.40 1.95 1 1.20 3 10 8.8 2.44 11 .518 .780
2 Ohio State 1.40 1.70 4 1.24 2 - - - - - -
3 Clemson 1.26 1.37 5 1.50 1 10 9.3 1.29 37 .787 .809
4 LSU 1.20 1.76 3 .98 6 10 8.3 2.41 12 .276 .552
5 Wisconsin .94 .97 6 .93 7 - - - - - -
6 Georgia .91 .80 9 1.15 4 10 7.4 2.26 13 .104 .206
7 Auburn .88 .58 22 1.12 5 10 6.7 2.84 3 .045 .126
8 Oklahoma .86 1.80 2 .06 52 9 7.1 1.45 34 .539 .406
9 Michigan .84 .64 17 .89 8 - - - - - -
10 Penn State .82 .75 12 .81 13 - - - - - -
11 Notre Dame .77 .73 16 .83 11 11 8.8 1.41 35 .184 .289
12 Florida .70 .58 21 .73 14 10 6.8 2.00 14 .050 .206
13 Washington .69 .50 31 .85 10 - - - - - -
14 Utah .68 .51 27 .82 12 - - - - - -
15 Oregon .64 .73 14 .58 18 - - - - - -
16 Iowa .63 .23 43 .87 9 - - - - - -
17 Texas A&M .52 .50 30 .55 21 10 5.2 2.81 4 .003 .006
18 Memphis .49 .94 7 .01 55 10 7.5 .86 47 .309 .047
19 Baylor .48 .43 34 .62 17 10 6.1 1.77 19 .089 .019
20 Minnesota .47 .73 15 .36 27 - - - - - -
21 Texas .46 .57 24 .35 30 10 6.3 1.60 30 .104 .119
22 Oklahoma State .44 .75 13 .16 43 10 5.9 1.77 20 .077 .059
23 USC .42 .76 11 .19 40 - - - - - -
24 Iowa State .40 .57 23 .24 39 10 5.8 1.63 28 .073 .068
25 TCU .37 .06 61 .65 15 10 5.7 1.61 29 .076 .067
26 Mississippi State .36 .52 26 .25 38 10 4.4 3.06 2 .000 .000
27 Navy .35 .88 8 -.08 62 11 7.7 .84 48 .275 .022
28 UCF .34 .58 20 .29 35 9 6.2 .74 52 .210 .041
29 Boise State .34 .39 35 .30 33 - - - - - -
30 Kansas State .34 .23 42 .35 29 10 5.4 1.85 16 .028 .011
31 Appalachian State .32 .19 44 .51 22 10 8.5 .25 75 .595 .001
32 Ole Miss .31 .51 28 .15 45 10 4.3 2.74 7 .003 .000
33 Washington State .31 .80 10 -.26 79 - - - - - -
34 Michigan State .31 .05 63 .63 16 - - - - - -
35 Virginia Tech .27 .12 52 .26 37 11 6.4 1.49 33 .010 .013
36 Air Force .27 .62 18 -.11 66 2 1.2 .19 77 - .001
37 North Carolina .26 .49 32 -.03 58 11 6.6 1.21 38 .011 .086
38 Missouri .22 .09 58 .36 28 10 3.9 2.80 5 .000 .000
39 Tennessee .21 .07 59 .39 26 10 4.0 2.59 9 .000 .000
40 Indiana .20 .53 25 -.11 67 - - - - - -
41 Kentucky .20 .17 47 .17 41 10 3.8 2.77 6 .000 .003
42 Florida State .19 .11 53 .40 25 10 4.8 1.76 21 .001 .004
43 Miami .19 -.14 80 .58 20 11 6.0 1.59 31 .001 .005
44 Arizona State .18 .07 60 .17 42 - - - - - -
45 Stanford .17 .28 40 -.08 61 - - - - - -
46 Cincinnati .16 .03 66 .33 32 9 5.4 .81 50 .082 .013
47 Virginia .16 .25 41 .01 56 10 4.9 1.65 27 .001 .005
48 South Carolina .15 -.17 82 .47 23 10 3.4 2.62 8 .000 .000
49 California .11 .10 57 .14 46 - - - - - -
50 Louisville .11 .58 19 -.29 83 11 5.6 1.29 36 .001 .007
51 Nebraska .09 .11 54 .05 53 - - - - - -
52 Pittsburgh .09 -.14 79 .34 31 10 4.5 1.73 24 .002 .004
53 West Virginia .09 -.02 70 .07 49 9 3.4 1.73 23 .007 .004
54 Texas Tech .08 .51 29 -.36 92 9 3.4 1.75 22 .008 .005
55 Florida Atlantic .08 .12 51 .10 48 9 6.3 .31 73 .178 .000
56 Houston .06 .18 46 -.14 70 10 5.1 1.12 41 .021 .000
57 Wake Forest .04 .03 64 .15 44 10 3.9 1.69 26 .001 .004
58 Northwestern .03 -.59 114 .58 19 - - - - - -
59 San Diego State .02 -.50 111 .44 24 - - - - - -
60 Duke .02 -.30 94 .26 36 11 5.5 1.19 39 .002 .007
61 SMU .01 .43 33 -.28 82 11 5.7 .97 44 .038 .000
62 UCLA .00 .15 49 -.16 72 - - - - - -
63 Western Kentucky .00 .01 68 .13 47 10 6.2 .48 58 .095 .000
64 Louisiana -.02 .32 36 -.36 91 10 6.5 .64 53 .200 .000
65 Purdue -.04 .11 55 -.30 84 - - - - - -
66 Boston College -.04 .05 62 -.07 60 11 4.6 1.81 18 .001 .000
67 Temple -.05 -.35 101 .30 34 8 3.6 1.01 43 .025 .008
68 Colorado -.06 .16 48 -.25 78 - - - - - -
69 BYU -.07 -.06 74 -.02 57 7 4.3 .43 64 - .000
70 Tulane -.07 .03 67 -.10 65 11 5.4 .92 46 .018 .000
71 Tulsa -.09 -.01 69 -.07 59 10 4.3 1.14 40 .017 .000
72 Wyoming -.09 -.33 98 .06 51 - - - - - -
73 Ohio -.10 .29 39 -.52 96 - - - - - -
74 Western Michigan -.11 .32 37 -.40 93 - - - - - -
75 Illinois -.13 -.37 106 -.10 64 - - - - - -
76 Syracuse -.15 -.18 85 -.18 74 11 4.3 1.84 17 .000 .000
77 Louisiana Tech -.17 -.12 78 -.08 63 10 6.5 .45 61 .366 .000
78 NC State -.18 -.18 84 -.19 76 11 4.4 1.05 42 .000 .000
79 Georgia Tech -.18 -.03 71 -.36 88 11 3.4 1.72 25 .000 .000
80 Georgia Southern -.18 -.36 104 -.15 71 10 5.8 .42 65 .104 .000
81 Oregon State -.18 .29 38 -.82 113 - - - - - -
82 Arkansas -.18 -.05 73 -.48 95 10 1.4 3.41 1 .000 .000
83 Army -.19 -.04 72 -.36 89 9 4.4 .62 54 - .000
84 Arizona -.19 .19 45 -.48 94 - - - - - -
85 South Florida -.20 -.34 99 .07 50 10 3.2 1.56 32 .004 .000
86 Buffalo -.20 -.20 86 .02 54 - - - - - -
87 Utah State -.22 -.32 97 -.17 73 - - - - - -
88 Marshall -.25 -.45 109 -.18 75 8 4.1 .40 66 .047 .000
89 Arkansas State -.25 -.22 88 -.24 77 11 5.5 .95 45 .049 .000
90 Maryland -.26 -.36 105 -.33 87 - - - - - -
91 Fresno State -.26 -.27 90 -.31 85 - - - - - -
92 Southern Mississippi -.27 -.35 103 -.14 69 11 6.7 .33 70 .159 .000
93 Vanderbilt -.30 -.35 102 -.31 86 10 1.6 2.51 10 .000 .000
94 Troy -.33 .03 65 -.67 107 11 5.7 .52 57 .032 .000
95 Middle Tennessee -.33 -.10 77 -.55 98 10 5.1 .32 71 .032 .000
96 Hawaii -.36 .14 50 -.91 117 - - - - - -
97 Miami (OH) -.37 -.65 116 -.11 68 - - - - - -
98 Toledo -.37 -.08 75 -.69 108 - - - - - -
99 Central Michigan -.38 -.29 93 -.36 90 - - - - - -
100 Colorado State -.39 -.22 87 -.63 105 - - - - - -
101 Eastern Michigan -.41 -.15 81 -.63 103 - - - - - -
102 UAB -.41 -.47 110 -.26 80 10 5.3 .38 68 .044 .000
103 Northern Illinois -.41 -.61 115 -.27 81 - - - - - -
104 Ball State -.45 -.27 92 -.56 99 - - - - - -
105 Kansas -.45 -.44 108 -.60 102 10 1.7 1.91 15 .000 .000
106 Florida International -.49 -.34 100 -.69 109 8 3.6 .27 74 .011 .000
107 Liberty -.50 -.27 91 -.60 101 8 2.9 .47 59 - .000
108 Kent State -.53 -.39 107 -.93 119 - - - - - -
109 UL Monroe -.54 .10 56 -.98 121 12 4.9 .46 60 .010 .000
110 Georgia State -.55 -.09 76 -.87 115 10 4.1 .43 63 .003 .000
111 San Jose State -.57 -.24 89 -.92 118 - - - - - -
112 Nevada -.58 -.66 117 -.56 100 - - - - - -
113 Coastal Carolina -.60 -.31 95 -1.00 123 10 3.8 .39 67 .004 .000
114 North Texas -.60 -.57 113 -.80 112 10 4.4 .31 72 .058 .000
115 East Carolina -.62 -.18 83 -1.09 125 9 1.7 .75 51 .000 .000
116 Charlotte -.65 -.32 96 -.98 122 11 3.3 .83 49 .003 .000
117 Rutgers -.67 -.92 125 -.64 106 - - - - - -
118 UNLV -.67 -.55 112 -.77 111 - - - - - -
119 South Alabama -.73 -.89 123 -.53 97 11 3.7 .35 69 .002 .000
120 New Mexico -.75 -.68 118 -1.01 124 - - - - - -
121 Old Dominion -.78 -1.01 126 -.63 104 - - - - - -
122 Rice -.81 -.91 124 -.88 116 8 2.9 .21 76 .006 .000
123 UTSA -.87 -.87 121 -.95 120 11 3.0 .56 56 .001 .000
124 Texas State -.90 -1.12 128 -.76 110 12 2.5 .57 55 .001 .000
125 Connecticut -.98 -.86 120 -1.25 128 - - - - - -
126 Bowling Green -.99 -.88 122 -1.22 127 - - - - - -
127 New Mexico State -1.00 -.81 119 -1.20 126 - - - - - -
128 Akron -1.16 -1.57 130 -.82 114 - - - - - -
129 UTEP -1.30 -1.15 129 -1.47 129 10 1.3 .44 62 .001 .000
130 Massachusetts -1.39 -1.08 127 -1.94 130 - - - - - -

Comments

2 comments, Last at 09 Sep 2020, 11:35pm

1 Given the exceedingly…

Given the exceedingly unbalanced number of games played per team, would it make more sense to track Expected Wins as Expected Win %, which at least normalizes the count a little?

Kansas and Air Force could end up with the same Expected Wins via entirely different means.

2 Nobody should feel the need…

Nobody should feel the need to apologize for not being sure how to predict the 2020 college football season; the most recent seasons that come close to rivaling 2020's weirdness is during World War II and the sudden appearance of service teams.  But I appreciate the concern.