This is an exciting time for those trying to find new and compelling ways to understand basketball through analytical means. The sphere of advanced statistical analytics is experiencing a golden era, and I was fortunate enough to be at ground zero this past weekend — the MIT Sloan Sports Analytics Conference. Some of the smartest people in the NBA world gathered in Cambridge to exchange ideas and pose larger questions about what we can — and in some cases, can’t — learn about the pro game from advanced data.
One of the things that continues to challenge people who work with this information is how to integrate individual data into a team game. If you’re a fan who grew up on Bill James Baseball Abstracts, then you have a fair understanding that baseball, at its essence, is an individual sport masquerading as a team sport, which makes it much easier to illuminate many of these questions. You can rationally measure how many runs a lineup with nine Dustin Pedroias would generate in 162 games vs. a lineup of nine Hanley Ramirezes, but basketball is a much tougher proposition. You could, theoretically, compare the offensive and defensive efficiency ratings of a team composed of five Chris Pauls vs. a team of five Yao Mings — and we have smart guys like Kevin Pelton on the case — but it wouldn’t produce results that are terribly useful in comparing the two players’ relative values. That’s because a direct statistical contrast between two individuals is a much trickier exercise in basketball.
Aside from determining how certain players are undervalued, what are some of the practical utilities for all this cool new data? One answer I heard repeatedly from the panelists and in conversations is applying this information to measure the performance of 5-man units. A couple of different stat guys told me that, for one, information about 5-man units can be imparted to coaches in a palatable way. An ornery coach might not want to hear from some data-cruncher that he needs to run fewer isolations sets for Player A out on the wing, but that same coach will likely be much more receptive to a few simple numbers that show he’s got a 5-man unit that’s killing the competition.
The Clippers are a tough nut to crack on a lot of this stuff because they haven’t been able to run consistent lineups out on the floor. The Clips have exactly one 5-man unit that’s shared the floor for more than 116 minutes this season: Baron-Gordon-Thornton-Randolph-Camby. That unit is a shade below average, with an adjusted +/- of -0.15, and an overall rating per 100 possessions of -2.17, which is a bit crummier.
Among the Clippers’ 5-man units that have played together for a measurable number of minutes, the best is Baron-Gordon-Collins-Randolph-Camby. In 84 possessions [about 46 minutes of basketball], this unit has outscored its opponents 98-72. Keep in mind that the standard error in such a small sample is pretty massive (Aaron Barzilai will tell you as much), and the majority of these data come from two games — the recent win over Boston and the December 12 Portland game. That being the case, these results support the notion that if you swap out Al Thornton for Mardy Collins on the wing, you get a substantially more efficient performance.
Naturally, the Clippers can’t capitalize on this information right now even if they wanted to, because Marcus Camby is suffering from head fluid, which sounds really unpleasant. But in thinking about how the Clippers might want to deal with LeBron James tonight, it’s hard to imagine throwing Al Thornton out there to guard him. I’m no fan of Mardy Collins’ inefficient offensive game. He displays a horrible habit of amplifying those inefficiencies by attempting far too many shots, a condition his coach needs to manage more vigilantly. That aside, the data show that on both an individual and team basis, Collins is considerably more useful than Al Thornton, who is the team’s least efficient regular, when placed alongside the Clippers’ other three primary scorers — Randolph, Gordon, and Baron Davis.
In some respects, this conversation evokes the debates of 2005-2006, when Corey Maggette and Quinton Ross were competing for playing time at the small forward position. Maggette was by every metric the more prolific offensive player, but there was enough data to suggest that the team played a more efficient game defensively when Ross was the SF alongside Cassell-Mobley-Brand-Kaman. The truth was that both Ross and Maggette had glaring deficiencies, but the ensuing discussion was one of the more interesting of its day for Clippers fans, with reasonable arguments on both sides.
Obviously, Thornton’s starting role hasn’t really been challenged by Collins, but in thinking about LeBron tonight…shouldn’t it be? If you had told me on Christmas Day that the matter of Collins v. Thornton would ever be a topic of earnest consideration, I would’ve eaten my hat, but among the dastardly number of issues that the Clippers are dealing with, the Al Thornton question stands out prominently. While I don’t support Al’s public humiliation by a tactless owner going off half-cocked, his role on the team needs to be examined more closely, particularly on a night when his team requires someone at the 3 who has some degree of defensive intuition. Again, Mardy Collins is not the long-term solution for the Clippers at the 3 — far from it — but most nights, he’s the better of two undesirable options.