Which is the best college football conference in 2017?

AnalysisAll analysis →

Every FBS team is a dot. Conference averages are not shown, and that is the point: the SEC’s mean is dragged up by four teams, the Big Ten’s spread is twice the Sun Belt’s, and a bar chart of means says none of it. What “compare the conferences” actually asks is how much of one league sits inside the other.

MetricCORE overallCORE offenseCORE defenseOffense EPA/playDefense EPA/play allowedOffense success rateOffense explosivenessPoints per drivePoints per drive allowedAverage starting field positionThird down conversionTeam talent composite

Season2026202520242023202220212020201920182017201620152014

CORE defense · 2017

One dot per team, one tick per conference median. Leagues are ordered by median, and the mean is deliberately not shown — four teams drag it and it hides how far the two distributions overlap.

All leagues ACC American Athletic Big 12 Big Ten Conference USA FBS Independents Mid-American Mountain West Pac-12 SEC Sun Belt

Conference	Teams	Median	Spread	Best	Worst
Big Ten	14	-8.4	15.5	Michigan -19.3	8.3
SEC	14	-5.6	16.8	Alabama -23.9	8.8
ACC	14	-4.8	11.8	Clemson -23.8	6.1
Big 12	10	-2.5	7.9	Texas -18.0	8.1
Pac-12	12	0.325	11.3	Washington State -11.9	13.7
FBS Independents	4	2.9	11.0	Notre Dame -10.9	10.7
Mid-American	12	4.0	7.5	Northern Illinois -9.3	18.1
Conference USA	14	4.1	9.4	Marshall -5.1	16.7
Sun Belt	12	4.9	8.4	Troy -6.3	21.4
American Athletic	12	5.7	13.0	South Florida -9.9	25.4
Mountain West	12	10.8	18.8	San Diego State -8.0	16.6

Two numbers, not one

Median answers “how good is this league”. Spread — the interquartile range — answers “how evenly good”, and it is usually the more interesting of the two. A league with a high median and a huge spread is four contenders and eight teams padding their records, which is exactly the shape that produces a conference champion who keeps losing in January.

Note which metric you are looking at. CORE is opponent-adjusted, so it can be compared across leagues honestly. The raw metrics cannot: a conference that plays itself twelve times has its schedule strength baked into every raw number, which is the single most common way conference comparisons go wrong.