One of the big stories of this year's Sloan Sports Analytic Conference, apart from the general struggle of trying to communicate complicated concepts in terms everyone can understand, was Kirk Goldsberry's presentation on new methods of measuring the impact and effectiveness of a player's interior defense. The paper got attention in part because the numbers and Goldsberry were heavily critical of Golden State Warriors All-Star David Lee, dubbed within as "the Golden Gate."
Lee has never been held in particularly high esteem as a defender, but the extent of the criticism was enough to bring the story into broader basketball circles. After Monday night's blowout win over the New York Knicks, Lee responded to the talk about his deficiencies. From Ethan Sherwood Strauss for WarriorsWorld:
Kirk Goldsberry’s Lee-skewering Sloan paper on interior defense, titled “The Dwight Effect,” came at the right time for Kirk and the wrong time for D-Lee. The Warriors were imploding defensively and Lee was getting plenty of blame. After the victory, I asked Lee for his thoughts on the paper:
“At this point I could care less. I’ve worked hard to improve my defense. I think I’m a much better defensive player today than I was a year ago and definitely to start my career. There’s a lot of different numbers to support a lot of different things. You can’t have it both ways. You can’t say me putting up 20 and 10 doesn’t matter because ‘numbers don’t matter,’ but at the same time, ‘charts at MIT matter.’ You can’t have it both ways.”
I’d say that’s a good answer under the awkward circumstances but not one I necessarily agree with. He’s a bit right that certain Lee critics castigate him for “empty stats” and some of those critics lept on these anti-Lee statistics with glee. But it’s not as though all Lee’s critics dismiss his offensive efficiency. The reasonable among us, I believe, praise his offense while deriding his D. For David Lee to escape the crosshairs of the next Kirk Goldsberry, he needs to improve beyond the personal defensive progression he cited.
Ethan makes two good points here: first, that criticism of Lee might have overwhelmed discussion because Goldsberry's presentation happened to coincide with a particularly bad stretch for the Warriors, and also that very few statistically inclined analysts don't credit Lee for his good offense even while terming his defense to be a massive spastic screwup.
Given the circumstances, this isn't a bad answer for Lee, even if he seems to give stats folk too little credit in their analyses of his overall game. There's no way Lee would ever bluntly state that he's bad defensively, but he also doesn't take the complete opposite approach and claim he's great. He simply says that he's improving, and that's not totally wrong.
Unfortunately, the premise of his statement is faulty. The analytics movement has never claimed to believe in all stats. To the contrary, it's always been about the idea that classic stats like total points and rebounds can misrepresent a player's contributions. The goal is to move towards better numbers, not accept them all and let people pick and choose depending on the point they're trying to make. Oddly, Lee has taken a common counterpoint for analysts facing of criticism — that tradition-minded basketball people do believe in stats, just the familiar ones — and turned it around to apply to a group of observers who pretty clearly don't want to accept all numbers. The goal of the movement has always been to identify better metrics, not to create new ones simply for the sake of convoluted computation.
Lee doesn't have to acquiesce to Goldsberry's findings entirely, and there's enough uncertainty surrounding the methodology of the study to not buy into every single finding as if it were gospel. Yet, if he wants to stand up for himself, there are ways to do it that don't disregard advanced stats entirely. Plenty of new metrics rate Lee well — it's even possible to argue that some overrate his impact. There are readily available ways to argue for his value to the Warriors without resorting to a tired, binary opposition between new and old stats.