In something of a surprise, Virginia’s game Saturday at Clemson was only its second this year against a projected NCAA Tournament team. It was also only its second against a team projected to play in either the NCAA Tournament or the NIT. Thanks to the coronavirus and some ACC scheduling quirks, the Cavaliers had, heading into Saturday, only played Gonzaga among teams projected to land in the field, with their other loss the infamous November defeat at the hands of NIT-bubble-adjacent San Francisco.
It wasn’t like Virginia had looked dominant in ACC play, too, even against the bottom end of the league. In a stretch in which they got to play Notre Dame twice and Boston College and Wake Forest each once, Tony Bennett’s guys never won by more than twelve. We knew they were good—our model bumped them up to ACC favorites after the last Notre Dame win—but we didn’t know they were capable of beating a team that had been playing as well as Clemson by 35 points, and on the road no less.
It was an eye-popping result, and for the first time in a long time, the ACC has a team legitimately in the title conversation (the fringe conversation, though—the one you get to after Gonzaga and Baylor). In today’s bracketology, UVA’s a three-seed.
Moving Up: Virginia, Minnesota
The other big one on Saturday was Minnesota’s pounding of Michigan in the Barn, and the Gophers rose accordingly, moving back onto the top four seed lines. Richard Pitino’s team is still projected to only barely finish above .500 in Big Ten play (this is a median projection), but it’ll be hard for them to get a legitimately bad loss the rest of the way, and their loss sheet so far—at Illinois, at Wisconsin, at Michigan, at Iowa—is strong, with home wins over Iowa, Michigan, Ohio State, and Saint Louis to bolster it. The Gophers aren’t much better than last year’s team, which finished ranked 27th in KenPom, but they’re getting the results this time, and against an even stronger Big Ten.
Moving Down: Clemson
Every action has an equal, opposite reaction…sometimes.
Michigan, you might note, did not drop from the one-line following Saturday’s loss. Clemson, meanwhile, dropped from a three-seed to a seven.
There are a couple reasons for the difference here: The first is that as teams approach Gonzaga and Baylor, their projected end-of-season résumés stratify, meaning the gap between, say, Iowa and Tennessee is wider than that between Alabama and Illinois, and by a sizable margin, despite each pairing landing on adjacent seed lines. The second is that the schedule environment in which teams’ projections live is not uniformly gradient. Sometimes, a reduction of one or two points per game in our model’s understanding of a team’s ability results in an added two or three losses to their projection. Sometimes, it doesn’t change the projection much at all. For Clemson, it was the former. For Michigan, it was the latter. This, too, has to do with stratification (you’d have to reduce Gonzaga’s ability rating by quite a few points to get a meaningful increase in projected losses, because Gonzaga is so much better than anyone left on their schedule), but it’s also just something of a quirk with the bracketology projection task itself: Teams have different environments. That’s how it goes.
Moving In: Michigan State, Western Kentucky (auto-bid), UC-Santa Barbara (auto-bid)
Michigan State played about exactly as expected Friday night, losing to Purdue in the final seconds in East Lansing. Their path to getting to that result wasn’t expected (the Spartans blew a large lead in the second half), but the final score was not a surprise.
Still, Sparty climbs back up.
Explaining why that happened:
First. Our model’s expectation of Michigan State’s final record, and therefore résumé, wasn’t significantly changed by the result. I don’t have the exact numbers on this (to increase the simulation count and thereby precision, we exclude some information in our readouts), but it’s possible the loss, being roughly 50% likely to begin with, didn’t move the final projected MSU W-L at all.
Second. The bubble picture is tight. Very tight. Michigan State’s median overall seed is 48th. So is Indiana’s. So is Florida’s. Loyola and Arkansas and Utah State check in at 50th. St. Bonaventure and Wichita State are at 51st. Marquette and VCU and Maryland are all at 52nd. Stanford’s at 47th. Boise State’s at 46th. San Diego State’s at 45th. Rutgers, Richmond, and Drake are at 44th. That’s sixteen teams within a seed line of Michigan State in their median projection. Small things can ripple a lot.
So, third. The small thing. While Michigan State’s result didn’t change their outlook, other results did. Again, we can’t pinpoint where exactly this happened, but the general picture seems to be that results like Illinois losing to Ohio State and Penn State losing to Purdue later in the weekend improved MSU’s regular season picture and probable conference tournament draws enough to bring them up just a hair. And a hair, in this case, was enough.
Elsewhere, Western Kentucky’s sweep of Marshall and UC-Santa Barbara’s strong performances against UC-San Diego vaulted the former ahead of Marshall in Conference USA projections and the latter ahead of UC-Irvine in Big West projections. At least for those respective tournaments (UC-Irvine might still be the regular season favorite).
Moving Out: Arkansas, Marshall (auto-bid), UC-Irvine (auto-bid)
Finally, in something simple, Arkansas’s failure to compete at Alabama pushed them across the cut line. The Hogs were expected to lose. They weren’t expected to lose that badly. And after showing a whole lot of life last weekend, Arkansas is now reeling after a rough week on the road.
***
The Friday/Monday cadence worked this week, so we’re going to tentatively plan on keeping it up for the next few weeks, potentially switching to Friday/Sunday/Wednesday as the calendar turns to February. In short, look for the next update Friday.
