The reaction yesterday to the first NCAA Evaluation Tool rankings for 2018-19 was so negative that there was a tendency in some quarters to see three-dimensional chess being played here by the powers that be in Indianapolis. Surely this is all a brilliant promotional strategy, a way to get otherwise preoccupied fans talking about college basketball in November.
By this reading, otherwise preoccupied fans could never have been induced to talk about college basketball in November by the release of rankings that were not the analytic equivalent of a drunken stevedore plummeting face-first down a manhole. Surely, the “needless ratiocinative mishap” aspect of the release was itself the very warp and woof of the next-level strategy at work.
Conversely, the release of sound rankings for all 353 teams all at once from a new system that will be used to actually select and seed the field for the 2019 NCAA tournament could never, by itself, furnish a coverage-worthy news peg. Got it.
For my part, I’m an avid enough reader of The Undoing Project to work under the assumption that attributing three-dimensional chess to an irredeemably opaque process likely says more about the observer doing the attributing and their efforts to attain congruence between priors and events than it does about the process itself. So be it.
We all have those priors, we all make those same efforts, and here’s where I’m at in my own quest for congruence….
Nobody knows anything
All commentary on the NET, present company included, is guesswork. The NCAA has reverted, strangely and surprisingly, to its Walter Byers roots and chosen not to share the blend of secret herbs and spices behind the NET.
Mark this decision as the NCAA’s first failure, and, make no mistake, the NET is above all else an administrative failure. We can have a rollicking good discussion of competing rating systems, and I know from personal experience there is no zeal like that of an analytically inclined individual to rip apart a new metric ushered forth by a large, entrenched, and lavishly funded bureaucracy. The cultural imperatives there are irresistible.
But before we get to the nuts and bolts, it bears noting at the outset that the opportunity cost incurred by the NCAA here is incalculable. Once Indianapolis had made a laudable if long overdue decision to kill the RPI, the slate was, for one hopeful instant, blank. Anything was possible, and it’s not at all clear that the NCAA even needed to invent something from the ground up in order to select and seed the field.
Nevertheless, we’ve been given not merely an entirely new metric but one cloaked in mystery. We don’t know how the NET would have sorted Division I on Selection Sunday last March, and we don’t know the relative weights given to the formulaic elements that the NCAA has seen fit to divulge. The RPI was a disaster, but at least it was a transparent disaster.
Not knowing anything is itself sufficient grounds to worry
It can be attitudinally satisfying to say that March NET, with the advantage of more games under its belt, won’t look as crazy as November NET. Additionally, saying this has the condign virtue of separating oneself from people who would have reacted negatively to the new metric no matter what, and there are always people who react negatively no matter what.
Still, in this case, expecting March to bring in its wake an analytic improvement is empirically sound but normatively and woefully insufficient. Indeed, if the NCAA really is brushed with some form of genius, it’s not for PR but instead for always setting forth these laughably low evaluative bars like “this will be better by March than what we saw yesterday.”
Pause and consider the corner we’re painted into, one where we’re using a needlessly opaque rating system to sort actual teams for the upcoming NCAA tournament. Of course, it could all turn out fine. We don’t know either way. That’s precisely the point. We don’t have to be in this position.
If we do know anything, it’s not terribly encouraging
Early efforts at reverse-engineering the NET are suggesting a rating system that badly wants to occupy a demilitarized zone halfway in between the RPI and KenPom. If that does indeed turn out to be the case, then we really are doomed and all those hysterical initial reactions will have been spot-on.
Creating a rating system that’s better than the RPI will prove to be a pyrrhic victory for the NCAA if the NET’s vulnerable to being gamed RPI-style by coaches seeking a profile edge, i.e., by all coaches. And early indications are that the NET could well be game-able.
Specifically, if the new metric’s as dependent on straight winning percentage and raw per-possession scoring margin as it appears, then we’re pretty much right back where we were with the RPI. The fact that a team’s NET score is closer to reality than the RPI used to be is a good thing in a vacuum, but neither the committee nor the public ever traffics in such raw scores. The coin of the realm is instead ordinal rankings, and if a coach can get a 0.01 raw score advantage by gaming the NET, that ranking will improve.
It will be wonderful if this does not turn out to be the case. It will be wonderful if the NET capably assesses how well teams play the sport. This was, after all, the promise of a new post-RPI dawn, to retire, once and for all, college basketball’s needless and irrational obsession with schedule at the expense of actual performance.
I am, as always, a hopeful meliorist, but anyone who says there’s nothing to worry about just because it’s November has it precisely backward. Neither the NET nor the process behind it have yet offered one affirmative reason not to worry.