How Good Is the SEC, Really?

You’re going to hear a lot about the SEC this week and how historic it is. This is true. The SEC is historic in terms of the number of teams it put in this NCAA Tournament and how those teams are advancing. It was also historically dominant in nonconference play, albeit only in the context of current scheduling norms.

Is this year’s SEC the greatest single-season conference ever? That’s a little subjective, but it’s a strong conference and it’s quite large. It might be. I’d love to see an analysis.

You might also hear a little bit about the SEC being over or underrated. The first round saw some SEC struggles and some accompanying SEC backlash. Accordingly, SEC haters finally came out of the woodwork to dunk on the SEC. The second round saw the opposite. The SEC didn’t struggle at all. Accordingly, it was SEC supporters returning from Homer Simpson’s hedge to dunk on the first round’s dunkers.

To be clear, this is all annoying. Fans shouldn’t cheer for their rivals. People who say things like, “[insert rival] winning means more money for us!” are correct, but they’re losers, and they’re mostly correct in a technical sense. (And that’s coming from a guy who built an NIT model.) For one thing, that money is marginal. For another, every other team in the conference is getting it too, which lessens the practical benefit. For a third, cheering for a certain team doesn’t make any difference in the outcome of the game. For a fourth, have a little self-respect. What these people really show—these Mississippi State fans repping the SEC, these Virginia Tech fans putting their hopes into Duke—is that some fans don’t believe in their own program’s ability to compete. They want to be carried. They want to live safely off the accomplishments of Baylor and Purdue. There are exceptions—as an Iowa State fan, I hold some sympathy for Kansas State and Oklahoma State, two schools similar in nature to ISU—but when a South Carolina fan wants Alabama to save their reputation…come on.

Revenue is not destiny in college sports. It’s helpful, but plenty of big-money teams lose and plenty of small-money teams win. This is truest in basketball, where we are only one year removed from Jack Gohlke, so can we please let the universe continue to play out before we declare mid-majordom dead?

We say all that because we’ve been tracking who’s overrated and who’s underrated this month, and we’re doing that at the conference level. I fear we’re feeding a beast here. So to be clear: We’re not worried about whether The Narrative is over or underrating a conference. We’re concerned with whether kenpom—college basketball’s closest thing to an answer key—is over or underrating a conference. Ken Pomeroy wrote a column pre-tournament explaining that his algorithm probably overrates the SEC, and then explaining why. The column makes clear that over or underrating in March should be correlated by conference, which is why we’re measuring it at the conference level. We want to know: Are Big 12 teams aggregately undervalued? Are SEC teams aggregately overvalued? That’s an angle here that isn’t often discussed, and it could be advantageous to our understanding of the sport. (And…yeah. Gambling.)

What we’re measuring is simple: What’s a conference’s record against the kenpom spread across NCAA Tournament and NIT games? More importantly, what’s its average point differential? Here are the numbers so far, limited to conferences whose teams have played five or more games:

The SEC did rise over the second round, and at a glance, the implication would be to not view the SEC as overrated. The problem is that the Big Ten, Big 12, and Big East are all overperforming expectations. You can quibble about the SEC being overrated or not, but if it isn’t, three of the other four power conferences being underrated gets you to the same effective place. Of course, there are caveats: This is a very small sample. If a conference played a lot of games against a conference kenpom overrates, that first conference might now be overrated by our analysis. Single teams can play a high share of a conference’s postseason games, like how Duke might play 40% of the ACC’s postseason when this is all said and done.

Other things standing out today:

• We publicly guessed that the ACC might be underrated heading into the tournament, willing to drink the kool-aid after deep runs by UNC and NC State and Clemson these last three years. We do still think it’s a tougher conference than the Mountain West—Kenpom says it is and Kenpom’s overestimated Mountain West teams by more points per game—but if there’s any meaning behind these numbers, it only points further towards a basketball Power Four that does not include Tobacco Road. (and the Bay Area, and Highland Park, and South Bend, and…)
• The more teams a league has in action, the more reflective these numbers will be for that league. This is relevant to the ACC, where we might end up in large part just measuring Duke, but it’s true more broadly as well. Going forward, these numbers will more and more reflect individual teams rather than broad swaths of conferences.
• The MVC, SoCon, AAC, A-10, and Big West all remain in play in the NIT, so we should get more mid-major data there. UC Irvine’s the favorite, but if the Big West really is overrated…

We’ve also been tracking over/unders, curious how inflated the ball is and isn’t. On Free Hoops yesterday, Greg Peterson offered a theory on why NIT totals have been so high. Go give that a listen if you’re curious.

In the NCAA Tournament so far, games are averaging totals 0.04 points higher than kenpom’s predictions. That doesn’t mean the ball isn’t overinflated, but it doesn’t support the claim that it is. Final scores have been very normal in their size, at least relative to the regular season. The median is a 0.5-point cover for the under.

In the NIT so far, the over is hitting by an average of 6.3 points. That’s a ton of points. The median is even higher: 9.5.

The sample’s small again here, and factors are dynamic and ever-changing. But most of the time, some information is better than none. This is the information we have.

**