Sunday, September 1, 2019

A look at some outliers

I've been keeping track of a bunch of offseason top 25s and such, and I have a fairly good handle on what the preseason conventional wisdom. Let's take a look at a some teams where the T-Rank projections diverge from the conventional wisdom, starting with teams that the T-Rank projections are much higher on than the humans.

Illinois
T-Rank: 15
Humans: 40-ish

The conventional wisdom on the Illini seems to be "intriguing, but not in the top 25." Andy Katz is more bullish, putting them at 19th in his latest "Power 36," but T-Rank's No. 15 is well above the median opinion.

I think this is driven by a few things.

First, Illinois returns some sophomores who were promising as freshmen, including a borderline one-and-done in Ayo Dosunmu. The model expects improvements out of those kinds of guys, and big improvement out of Ayo.

Second, the model still likes Brad Underwood based on his well-above-par performances at Stephen F. Austin and Oklahoma St. Humans are not so sold on Underwood, and there's quite a bit of skepticism that his particular system, especially on defense, can work at the high major level.

Third, T-Rank thought Illinois was better last year than most people (and other systems like Kenpom) did. This can have an effect on the defensive projection, which uses prior performance as a baseline.

My take. This is definitely a best-case scenario projection for Illinois, and I'd be much more comfortable with them in the lower 20s even as an optimistic case. There are things to like about this team, and plausible bases to expect a significant leap, but still many question marks and things that could easily go wrong. So I'd go with the conventional wisdom on this one.

Missouri
T-Rank: 13
Humans: Bubble team at best

This is definitely the biggest outlier in the projected top 40. As far as I can tell, no human is expecting anything out of this Missouri squad. Only mention I've seen is they come up 64th in Kevin Sweeney's top 100.

The case against Missouri is straightforward. They were mediocre last year (68th in Kenpom), they lose their most impactful player (Jordan Geist), and they don't have a highly touted freshman class coming in.

As with Illinois, my model's baseline for Missouri is a bit higher on Missouri's past performance; T-Rank had them 54th last year, which is higher than anyone else had them. Most importantly, it had them No. 36 in adjusted defensive efficiency, and 37th the year before. Add in the fact that Cuonzo Martin has a history as a solid defensive coach and that Missouri returns a solid core of defensive players, and I don't have a big problem with the projection for Missouri to be the 25th best defensive team next year.

Where things get a bit wacky is the projection for Missouri to have the No. 15 offense. The offensive projection is mainly player-based, and the short story here is that the model just likes these players together. Mark Smith and Jeremiah Tilmon are juniors who were highly rated recruits, which bodes well for their chance to continue improving into key, reliable offensive players. Torrence Watson, Xavier Pinson, and Javon Pickett are returning sophomores who got good experience as freshmen. And Dru Smith is a transfer from Evansville (a decent mid-major) who was very efficient there as a sophomore.

My take. As with Illinois, I'd find this more defensible if they were in the low 20s or so, because I can sort of get where the model is coming from in liking these players. But I'm not placing or taking any bets premised on Missouri as a top-20 team that contends for an SEC title.

Penn St.
T-Rank: 19
Humans: 50-ish

Penn St. has been a pretty decent team, efficiency-wise, the last two years. Two years ago, they won the NIT and finished 19th in the Kenpom ratings. Last year they finished 43rd at Kenpom. (T-Rank had them 16th and 32nd the past two years, so even a little higher.)

But the perception of this team does not quite track those efficiency numbers. They were never seriously in consideration for an at-large berth either year, mainly because of their 6-20 record in "Quadrant 1" games. Starting 0-10 in the Big Ten last year also didn't help. There's a feeling, I think, that PSU's efficiency numbers are goosed by mopping up against lesser competition, and that Pat Chambers just doesn't have the chops to really compete at the top of the Big Ten.

So the popular opinion on Penn State, I think, is they they were mediocre with a low ceiling last year, and they'll be similar this year after losing Josh Reeves and the unexpected transfer of Rasir Bolton. Yes, they return Lamar Stevens, who puts up big counting stats but does so inefficiently. So it's easy to imagine this being the classic PSU team that features one guy who takes all the shots and that plays pretty hard enough on defense.

My take. 19 is too high, but it's worth noting that Penn State was 24th in a sneak peak of Kenpom's preseason ratings, so at least T-Rank isn't out on an island with this one. Maybe this will be the year that PSU's results more closely align with their efficiency numbers.

Notre Dame
T-Rank: 22
Humans: 50-ish

I've seen surprisingly little love for Notre Dame this offseason. The Irish struggled last year, especially after Rex Pflueger got hurt after ten games. Three top-100 freshmen (Nate Laszewski, Dane Goodwin, and Prentice Hubb) were forced into more action than they were probably ready for.

But Pflueger is back for a fifth year, along with fellow seniors TJ Gibbs and John Mooney. And those three top-100 freshmen are back, poised to make the sophomore leap.

My take. Notre Dame has been a bit snakebitten since Bonzie Colson's injury in 2018. If they stay healthy this year, I think there's a good chance that they are a top 25 team again.