Is Duke the least consistent really good team in the country, or does it just seem that way?
It's a fair question to ask after a near-perfect performance against Virginia Tech was followed by a double-overtime loss at Wake Forest in a game Duke was favored by double digits, and preceded by the absolute clunker at N.C. State, Duke's worst performance of the season.
When Duke is good, the Blue Devils might be the best team in the country. When they're bad, they can lose to anyone, even a mid-major opponent at home for the first time in 37 years. (Stephen F. Austin, it should be noted, is worthy of an at-large berth into the NCAA Tournament, even if it won't get one if it comes to that.)
The ups and downs are enough of a yo-yo to give anyone whiplash, and the numbers bear out the crazy swings in performance. But are those swings really that different than any other team? Like, say, Virginia, where Duke has to travel Saturday?
With help from college basketball stat guru Bart Torvik, it turns out that they are, at least a little bit.
Torvik's "game score" measures a team's performance against its expected performance against that opponent on a 0-100 scale. Duke's wins over Notre Dame (by 34) and Virginia Tech (by 22) were both 99s. The State loss was a season-worst 20; the Wake Forest loss was a 64, third worst after the Wolfpack and the Lumberjacks.
The graph of Duke's game scores looks like a Chutes and Ladders board.
But looking at only Duke doesn't actually tell us much.
By comparing the variance in Duke's game score over the course of the entire season to the rest of Division I, Duke is actually one of the more consistent teams in the country, 31st out of 354. But when measured only against their peers, the top 20 teams in Ken Pomeroy's efficiency ratings, the Blue Devils are 12th in terms of variance.
Against the true elite, the KenPom top 10, the legitimate national title contenders, Duke is eighth, ahead of West Virginia and Michigan State.
All of which is to say, the Blue Devils aren't any less consistent than the average team, but they're a lot more inconsistent than the really good ones.
There's isn't a simple, obvious reason why. It's not an over-reliance on the 3-pointer, sinking and swimming from long range. This Duke team relies less on the outside shot than all but one in the past two decades. Only the 2007 team _ Jon Scheyer's freshman year _ has taken fewer than this one.
It's not point guard, where Tre Jones has been an uncommon luxury as a sophomore, both as a ballhandler and an on-the-ball defender. Unlike the 2018 group, this Duke team hasn't been forced to play zone out of necessity.
And it isn't even a young team by Duke's standards. It just plays young at the wrong times, still figuring things out at this late date in the season, as good as the Blue Devils have been at times. It's more the combination of the same youth without the same kind of talent Duke has had the past two seasons that's causing the problems, managing the inevitable grind of the conference schedule.
"There are many more interruptions that produce a variance in performance, or can produce a variance, in our sport," Duke coach Mike Krzyzewski said. "You just try to do your best every day and hopefully that best translates to when you're playing a game."
There's some good news in this for Duke. For one, regular-season inconsistency does not necessarily lead to postseason inconsistency. Torvik has not found the game scores to be predictive of future performance. But high-variance teams in general do tend to boom or bust in the postseason, a scary thought for a program that has lost three times in the opening weekend of the NCAA Tournament in North or South Carolina in the past eight years and is ticketed for Greensboro in the first two rounds.
On the boom side, though, there's the example of Connecticut in 2014. The Huskies bumbled and stumbled through the regular season, only to get a ridiculously smooth path through Buffalo and Madison Square Garden as a No. 7 seed and win the national title. UConn was really bad at times but unbelievably good at times, and only the unbelievably good showed up in March (and April).
Duke's resume to this point is much better than UConn's was, but the ups and downs are familiar, and there's a decent chance Duke, like UConn, could play de facto home games in Greensboro and the Garden if it can finish strong and stay among the top five or six overall seeds. It's just hard to know what to expect from the Blue Devils, on Saturday and beyond.
