Keep Winning, Alabama, And You May Be The Best Team In History

There will never be a definitive ranking of the most dominant college football teams of all time — and in a sport in which the subjective act of voting was the primary mode of determining a champion until only very recently, that’s probably fitting. Even the computer power ranking systems can’t always agree (as the Bowl Championship Series found out time and again). But according to one particular metric we like to use here at FiveThirtyEight — the Elo rating — the 2016 Alabama Crimson Tide are closing in on the greatest college football season in at least 80 years.

Here are the rankings as they stand now, using the Elo rating system I developed with FiveThirtyEight editor-in-chief Nate Silver during the 2014 season:

The best college football Elo peaks (1936-2016)
	SEASON	SCHOOL	DATE OF PEAK	RECORD AT PEAK	PEAK ELO
1	1995	Nebraska	1/2/96	12-0	+33.9
2	2016	Alabama	11/12/16	10-0	+33.4
3	2015	Alabama	1/11/16	14-1	+32.8
4	1974	Oklahoma	11/30/74	11-0	+32.3
5	1988	Miami (FL)	1/2/89	11-1	+32.2
6	1973	Oklahoma	12/1/73	10-0-1	+31.9
7	2008	Florida	1/8/09	13-1	+31.9
8	2012	Alabama	1/7/13	13-1	+31.9
9	2004	USC	1/4/05	13-0	+31.8
10	2014	Oregon	1/1/15	13-1	+31.6
11	1971	Nebraska	1/1/72	13-0	+31.6
12	2014	Ohio State	1/12/15	14-1	+31.6
13	1997	Nebraska	1/2/98	13-0	+31.3
14	2005	USC	12/3/05	12-0	+31.2
15	1996	Florida	1/2/97	12-1	+31.0
16	2002	Miami (FL)	12/7/02	12-0	+30.9
17	2013	Alabama	11/23/13	11-0	+30.7
18	2011	Alabama	1/9/12	12-1	+30.4
19	1945	Army	12/1/45	9-0	+30.3
20	1989	Miami (FL)	1/1/90	11-1	+30.1
21	2009	Florida	11/28/09	12-0	+30.1
22	1975	Oklahoma	11/1/75	8-0	+29.9
23	1993	Florida State	1/1/94	12-1	+29.8
24	1973	Alabama	12/1/73	11-0	+29.8
25	2011	LSU	12/3/11	13-0	+29.7
26	1996	Nebraska	11/29/96	10-1	+29.7
27	2001	Miami (FL)	1/3/02	12-0	+29.6
28	1983	Nebraska	11/26/83	12-0	+29.5
29	1992	Miami (FL)	11/28/92	11-0	+29.5
30	1999	Florida State	1/4/00	12-0	+29.4

Some additional background on the rating system: It’s designed to mimic ESPN’s Football Power Index, in the sense that it’s optimized to predict future results, and it’s on the same scale as points per game (so you can take the difference between any two teams’ ratings and get a basic point spread if they played at a neutral field).¹ Just like in our NFL Elo system, the ratings update after each game depending on who won, by how much and where the game was played, and before each season, the ratings from the year before are carried over after being reverted slightly toward the mean.²

The upshot is a rolling power rating for each team that tries to pinpoint how good they are at any given moment. And at this particular moment, Alabama is really, really good. After their 51-3 destruction of Mississippi State last Saturday, the Crimson Tide’s Elo rating moved to +33.4 — meaning they grade as about 34 points per game better than the average FBS team.³ (Think of Air Force, South Carolina or Kentucky.⁴) That’s the second-highest peak rating recorded by any Division I-A/FBS team going back to the start of the AP poll era in 1936, trailing only Nebraska’s +33.9 rating after the Cornhuskers crushed Florida in the 1996 Fiesta Bowl.

Alabama won’t get a chance to break Nebraska’s record this weekend. The Tide do have a game, but it’s against Chattanooga, a Football Championship Subdivision opponent, and the game is so lopsided that Elo will give ’Bama barely any credit for a win, no matter the margin. (Within reason, of course: Alabama could technically pass Nebraska if it beats Chattanooga by 28,228 points — or about a touchdown and a two-point conversion every second of the game. Eat your heart out, 1916 Georgia Tech Yellow Jackets.) But assuming that its rating stays constant against the Mocs, Alabama could surpass Nebraska’s rating if it beats Auburn by at least 32 points a week from this Saturday, in the regular-season finale. That’s not likely (Elo considers Alabama only 20-point favorites), but because Elo guarantees a ratings boost for wins against good competition, the Tide would almost certainly break the record if they beat Auburn by any margin and maintain their winning streak through the SEC Championship Game and the College Football Playoff.

Will that make them the greatest team of all time? Like most things in college football, that’s up for debate. But it is one piece of evidence from which they can build a case. And it’s bad news for any opponent standing between them and coach Nick Saban’s sixth career national championship.

Footnotes

Our College Football Playoff model also has its own Elo component, but that version is used to predict how the CFP committee will react to results. Confusing, I know.
The particulars of college football are slightly different than in the pros, of course: College ratings carry over much more from season to season — since programs tend to stay relatively stable over time — but the amount by which each game changes the ratings (in Elo parlance, the “K-factor”) is much larger in college. (College teams play fewer meaningful games in a season, but when they do, we tend to learn a lot about each side.) Also, all non-Football Bowl Subdivision teams are assigned a common (very low) rating based on how well that type of team did against FBS teams over the last few seasons. But just like in the NFL, when a team plays its first season in the FBS (or hasn’t been classified there in more than three seasons), they are assigned a low rating, and the offseason reversion process moves them upward toward a slightly-above-average rating each season they play.
As of Thursday’s games, the average Elo across FBS is -0.5.
As a sanity check on Elo, when Alabama played Kentucky earlier in the season in Tuscaloosa, the Tide were favored by 37.5 points, which works out to a spread of about 34 at a neutral site. (’Bama won by 28.)

Footnotes

Comments