Interesting article regarding conference strength (Kenpom)

[The_Answer1313]

Something I found a bit interesting given the talk about conference strength and what that might mean for the tournament and whatnot. Sorry for the bad format, this comes through a newsletter in the email. Still worth a read. Why hasn't anyone said The SEC is overrated. I’m not sure why people care so much about which conference is best. But if you remember last season, there was an unusually organized sub-culture of fans and analysts that criticized the Big 12 for being overrated and “gaming the system” in the weeks leading up to the tournament. It’s kind of strange that no similar backlash has developed against this season’s SEC, a league that seems to have unanimous support as not only this season’s best conference, but the best ever. (The reason there isn’t the same backlash is because most of the Big 12 haters were ACC boosters and the ACC is even more down this year, although it does potentially have the pre-tourney favorite to win the title so all I’m saying is that it’s not too late to lean into it again.) Indeed, based on non-conference play, the SEC has been one of the best leagues of the internet era, and therefore, possibly ever. The conference produced the best non-conference winning percentage over that time and the best scoring margin since the 1997 ACC: ​ The ‘97 ACC is worth remembering, though. Theoretically, it was the best conference of modern era, and yet its tourney performance that season was sad-trombone worthy: North Carolina (1 seed): lost in Final Four Duke (2 seed): lost in second round Wake Forest (3 seed): lost in second round Clemson (4 seed): lost in Sweet 16 Maryland (5 seed): lost in first round Virginia (9 seed): lost in first round The 2024 Big 12 is third on the list in terms of non-conference scoring margin and you may remember how their tournament turned out because the haters were vindicated. To refresh your memory: Houston (1 seed): lost in Sweet 16 Iowa State (2 seed): lost in Sweet 16 Baylor (3 seed): lost in second round Kansas (4 seed): lost in second round Texas Tech (6 seed): lost in first round BYU (6 seed): lost in first round Texas (7 seed): lost in second round TCU (9 seed): lost in first round Collectively, the two conferences with the best non-conference scoring margin in history sent 14 teams to the tourney and got one team to the Elite Eight. One takeaway is that it’s not unusual for historically great conferences to fail in the tournament. So maybe it’s time to sell the SEC. Before Greg Sankey throws a fit, it’s important to note I’m not disputing the SEC’s conference supremacy this season. Just that it’s likely the SEC’s tournament performance will not be entirely consistent with being the best conference of all-time. We’ll get to the data, but I think there’s a strong non-data argument here about why a conference with extremely strong non-conference numbers is set up for tournament disappointment. If you’re one of them bayesian types, there are two ways to think about it. Any group of teams that collectively exceeds expectation at anything in November and December is headed for some regression. Whether it’s 3-point shooting, offensive rebounding, or just general performance, this is always going to be true. If we had another two months of non-conference action, the SEC would still be really good but not quite as good as they looked earlier this season. We really have no idea if conferences as a whole are improving or not after about January 1, when non-conference play ends. And the talent difference among power conferences probably isn’t as great as the difference in performance in non-conference play would indicate. This season, ACC teams have had more room to improve in general, while it’s hard to believe that SEC teams as a whole could actually get better since non-conference play ended. But it’s really difficult to assess that and to my knowledge, nobody actually tries at all. (Which is another piece of evidence as to why the “eye test” is entirely bogus. If people were so good with the eye test, you’d hear things like “the SEC has really dropped off since conference play started”, but that never happens¹.) There’s probably more influence from (1) than (2) but both could be factors. And if you really want to see data, it’s consistent with this theory. Going back to 1997, I took the pre-conference rating of every power league - relative to all power leagues that season² - and plotted that against the difference in their pre- and post-tournament rating. The trend is clear: ​ Among power conferences, the better the league, the more likely their rating will drop during the tournament. I’ve added the linear trend, but it’s not certain that the trend is linear. The very best conferences have underperformed even more than the linear trend would indicate. What does this all mean? It’s another example of how preseason priors are still useful in informing predictions late in the season.³ I think by March we have a pretty good idea about how teams stack up relative to others within a conference. You can get away without any team priors at this point in this season. However, there’s not enough in-season data to know precisely where conferences stack up relative to each other. We basically have zero information on that front since January 1st. So you can amaze your friends by predicting that conferences that did extremely well in November and December will revert to the mean during postseason play. You’ll probably be right. 1 With notable exceptions from Archie Miller and Rece Davis this season. But in neither case were those gentlemen using some sort of subjective evaluation. They were badly misunderstanding some data. 2 It’s not the point of this piece, but while the SEC has the best conference rating in modern history, both the 1997 and 2004 ACC had more impressive ratings relative to other power leagues. The 2004 ACC had a rating that was over seven points better than the next-best league. This year’s SEC is a mere four points better than the second-best. 3 It’s not the point of this piece, either, but if the NET rating is the baseline for resume metrics, and SEC teams’ ratings are slightly inflated, then the league’s teams are benefitting from slightly inflated resumes.​