Last March at the Sloan Conference in Boston, I was told a near-perfect parable on the traditionally and deeply yet needlessly antagonistic relationship between the “talent” and “analytic” schools of basketball interpretation.
In the immediate aftermath of a dismal 2013-14 season, an NBA general manager ordered a top-to-bottom review of what had gone wrong with the team that year. By that time every front office was fully equipped with bright young minds who could apply the latest analytic tools and even brandish some proprietary and closely-held statistical methods of their own. But the GM had allotted just 48 hours for the task while also imposing a draconian two-page limit on the final report. As a result the analytics team worked in a frenzy to summarize every last data point, shot chart, and pick-and-roll efficiency in just a couple of pages.
At the end of the ordeal the exhausted head of the analytics group yanked the final draft from the printer and thrust the two-page encyclical into the waiting hands of his boss. Whereupon the GM took the report and, smiling genially and never so much as glancing down at the printout, wadded up the two pages while taking aim at the nearest waste basket. As the GM let fly with his shot, he uttered one word:
After a decade of watching college basketball in the one-and-done era, I’ve come to the conclusion that in one crucial respect the GM is exactly right. In fact the more I ponder the question the more I think I’ve become something of a talent essentialist in spite of myself.
I wonder whether there might be rare instances where we can remove college performance from the equation more or less entirely and just look at the roster of players. Forget efficiency or shooting accuracy. Tell me how many minutes the returning players recorded, how many possessions they used, and how highly the freshman class is rated, and in these very rare instances this may be all we need to know.
In such cases I think we may be able to just look at a college basketball roster before the season even starts and say that if precedent’s any guide this team has virtually a 100 percent likelihood of earning an NCAA tournament No. 1 seed, a 60 percent probability of reaching the Final Four, and a two-in-five shot at winning a national title. I’m going to call such instances category 5 rosters, and, though I (and others) didn’t know it ahead of time, it turns out that Duke’s in 2015 was one such roster.
The baffling analytic mystery posed by John Calipari’s existence
The NBA has always descended visigoth-like upon the college game and harvested its very best players annually, but what changed on July 29, 2005 is that these players are now being removed from the college scene after a single season in Division I. Once this policy had been instituted it was only a matter of time until a college coach realized that an unprecedented and rather amazing opportunity was now at hand. In a one-and-done world it’s possible to transcend the traditional conflict between recruiting the best players and finding roster spots for all your talent. For the first time in the history of college basketball, it’s possible to have the best recruiting class in the nation every single season.
That John Calipari would be the coach to seize this opportunity now seems inevitable, but it very easily could have played out in any one of a number of different ways. Actually for a heartbeat in the late aughts it appeared that it might be Roy Williams, Thad Matta, or even Ben Howland who would carry this torch. And if the previously excellent Billy Gillispie hadn’t inexplicably gone all Colonel Kurtz during his tenure at Kentucky, who knows what the college hoops landscape would look like today.
Be that as it may, Calipari was hired by Kentucky on April 1, 2009. He promptly signed John Wall and DeMarcus Cousins, and the rest is history. In every season since Calipari’s arrival, UK has brought in a freshman class that, by the standards of Drew Cannon’s canonical front-loaded recruiting curve, contains at least 25 recruiting points. How good is that?
Freshman classes with 25 recruiting points or more 2006-15 Kentucky 7 (2009-15) Duke 2 (2014 & 15) UCLA 2 (2008 & 12) North Carolina 2 (2006 & 09) Kansas 1 (2013) Ohio State 1 (2006) Years refer to recruiting classes: "2015" freshmen arrived this fall
In the entire one-and-done era there have been just 15 such 25-point classes nationwide. Calipari alone has accounted for seven of them.
The empirically convenient part of all this is that Calipari has signed these classes consecutively and without interruption. Indeed you can look at college basketball for a very long while and not find anything as consistent as “John Calipari’s recruiting.” Still, that very consistency poses something of an analytic problem and suggests an obvious question for the NBA general manager quoted above. If players really are everything and one grants that Calipari signs as much or more talent than anyone else annually, why has his recruiting been so much more consistent than his NCAA tournament seeds?
Kentucky's NCAA seeds under Calipari 2010 1 2011 4 2012 1 2013 NIT 2014 8 2015 1
Just because you recruit really well doesn’t mean you’re impervious to bad luck, of course, and certainly the NCAA’s eligibility ruling on Enes Kanter in 2010-11 and the injury suffered by Nerlens Noel in 2013 both influenced these numbers. Even making due allowance for such factors, however (in 2012-13 UK was a respectable but by no means scary 17-6 with Noel), it’s clear that success for Kentucky has required more than simply another amazing freshman class. After all, Calipari always has another amazing freshman class. So why does that translate into top seeds sometimes but not others?
Talent, experience, and redundancy
Maybe having the best freshmen in the country is tremendously advantageous but not sufficient in and of itself. Maybe returning experience matters too.
Actually, if the team of interest is Kentucky there’s no “maybe” about it. Returning experience matters, period. Here are all six of Calipari’s UK teams, listed from least to most returning experience (as measured by returning possession-minutes or RPMs, in effect each returning player’s percentage of possessions used multiplied by that of minutes played). See if you spot a trend:
%RPMs SEC W-L Seed 2013 6 12-6 NIT 2011 12 10-6 4 2014 30 12-6 8 2012 53 16-0 1 2010 55 14-2 1 2015 60 18-0 1
In other words if you’re Calipari and you can be trusted to cancel out the “recruiting” portion of this equation by holding it constant annually, returning experience is fate. So the challenge for the rest of D-I’s blue chip programs would seem to be to recruit as well as Calipari, even if it’s just once, and bring back some talent that same season. Thus the definition:
A category 5 roster is one that returns at least 40 percent of the previous season’s possession-minutes and adds a freshman class with at least 25 recruiting points.
There have been five such rosters in the one-and-done era. All five teams were NCAA tournament No. 1 seeds:
- North Carolina in 2007 (Tyler Hansbrough and Ty Lawson: 31-7, Elite Eight)
- Kentucky in 2010 (John Wall and DeMarcus Cousins: 35-3, Elite Eight)
- Kentucky in 2012 (Anthony Davis and Michael Kidd-Gilchrist: 38-2, national title)
- Kentucky in 2015 (Karl-Anthony Towns and Willie Cauley-Stein: 38-1, Final Four)
- Duke in 2015 (Jahlil Okafor, Justise Winslow, and Tyus Jones: 35-5, national title)
Three things strike me about this list.
First, Kentucky 2014 isn’t here. You may remember that was the team (featuring Julius Randle, James Young and the Harrison brothers) that occasioned “40-0” shirts in the preseason, was labeled a huge disappointment by February, and then made the national championship game anyway as a No. 8 seed before losing to Connecticut. We will be waiting a very long time before we see another recruiting class like that one, but the Wildcats were lacking that season in the “returning experience” department.
Second, there’s a saying in economics to the effect that recessions and depressions uncover what the auditors missed. In somewhat the same fashion maybe the redundancy supplied by veterans compensates for any mistakes made by recruiting analysts. Having returnees on hand can offset the rare but nevertheless recurring instances where even freshmen rated in the top five nationally fail to deliver on expectations. Under this reading college basketball at the very highest level (currently Calipari and Mike Krzyzewski, full stop) resolves itself into a challenge of talent acquisition and sequencing. That process — comprised of such granules of tedium as in-home visits, texts, campus visits, etc. — is intrinsically boring and more or less impervious to spectating but may hold a far greater influence over actual team performance than all of the things (player development, so-called bench coaching, etc.) that we write and tweet about so lovingly.
Lastly, Wisconsin in 2015 was one of the unluckiest teams I’ve ever seen, at least in terms of timing. The Badgers had to face two of the greatest rosters of the entire decade within the span of 48 hours in Indianapolis. Give Bo Ryan major credit for going .500 in that particular two-game swing.
Of course it’s hardly a shocking research breakthrough to note that if you’re somehow able to combine a Calipari-level freshman class with some returning experience you’re going to have an excellent shot at a national title. It’s just that lately I’ve found myself trying to define somewhat more precisely the truisms we already know and mouth to each other. I do think that if I’d had this definition handy a year ago I would have looked at Duke in a different light. Specifically I would have expended less real-time analytic energy fretting about the Blue Devils’ defense.
A category 5 forecast for 2016
This won’t take long. There are no category 5 rosters this season. The freshman classes at Duke and Kentucky meet the criteria, but the levels of returning experience on these two rosters do not.
%RPMs Recruiting pts Duke 2016 24 27.6 Kentucky 2016 16 26.9
Maybe these numbers somewhat understate the potential strength of both teams. Presumably UK would show a better figure for returning experience had Alex Poythress merely been healthy all season in 2014-15. And there’s a case to be made that Grayson Allen’s minutes as a freshman weren’t what one will typically see from a player that stands a reasonable chance of being a first-round pick after his sophomore season.
Then again we can hang similar “yes, but” asterisks on other teams besides Duke and Kentucky in 2016. Surely it’s notable as well that this season the rest of D-I will, apparently, have some blue sky above them in the form of a category 5 absence. In recent seasons when there are no such rosters, we’ve seen not only a top seed but also a No. 3 or even a No. 7 seed crowned as national champion. Other things being equal it’s possible there’s a higher than customary likelihood of such a dark horse emerging once again next April.