Last Friday, NBA.com made SportVU player tracking data available to the public for the first time. It’s just the latest development in the fast-moving basketball analytics revolution, but it portends a massive shift, not only in how we watch games, but also how we value players, rate coaches and evaluate front office decisions. Using cameras perched high above the floors of all 29 NBA arenas, STATS LLC has already influenced how teams make such determinations, but now, for better or worse, we fans get an invite to the party, too.
Predictably, with change comes ideological conflict. The basketball analytics movement pits old-school eye testers versus new school statisticians. Reasonable people seek the middle ground, and rightly so. What my-own-two-eyes tell me (“MOTE”) and advanced analytics each lend a distinct perspective on the game, but neither should be treated as definitive and neither should be ignored. Instead, SportVU, which tracks player movement and distills same into useful data, is that middle ground. (I wrote about four of the new SportVU categories over at NBA.com, but there are actually nine total metrics available. You can check them out here.)
Knicks fans make for a fascinating case study in connection with the stats v. MOTE debate. The New York fan-base vacillates daily on which lineups coach Woodson should play, just how valuable the club’s players are. Take the Andrea Bargnani acquisition, for example. That deal was generally lambasted by the analytics crowd, but lauded by eye-testers. Perhaps more than most of his peers, Bargnani is a quintessential flashpoint player for these kinds of discussions. Part of it stems from the fact that he is a seven-footer with unique offensive skills. Part of it is because he was drafted number one overall by the Toronto Raptors. Those who rely upon MOTE point to what Bargnani can do on a basketball court, in theory. He is a versatile pick-and-pop big who can shoot from a variety of spots on the court. Stats folks, however, will tell you that Bargnani has been a well below average player over past four seasons by most any statistical measure.
While the system won’t conclusively end every such debate — whether we’re arguing about Bargnani or anyone else — it can provide context and a relative value of the data that is generated. By measuring certain aspects of the game, including, but not limited to, catch-and-shoot jumpers or rim protection, we can begin to determine if there is a correlation between how we view players versus what the raw numbers suggest about those same players. Unfortunately, the sample size is ridiculously small at this point, so there aren’t many, if any, meaningful conclusions to be drawn. And though the currently available data is not expansive – SportVU has released only a handful of the data it collects - we can glean some information about the Knicks team as presently constituted.
In the world of basketball analytics, the advancement of good defensive metrics lags far behind what has been accomplished in evaluating offense. Defense, of course, tends to be more team-oriented, and as such, it is difficult to quantitatively measure players on an individual basis. SportVU does track basic defensive actions like blocks and steals, but it also measures defensive rim protection — defined as an opponent’s FG% when said defensive player is within five feet of both the basket and the opponent.
In a perfect world, the aforementioned metric would have been available for each of the last two seasons. For example, how did Tyson Chandler’s rim protection numbers differ? MOTE tells us that he was a much better in 2011 — he won the Defensive Player of the Year Award – and the numbers were certainly down in his injury-riddled 2012 campaign, but to what extent did his defense actually suffer versus what our perception was? Through the first three games of this young season, Chandler “looks” closer to peak-form than I’m-getting-abused-by-Roy-Hibbert-please-get-me-outta-here form. Do the numbers bear that out, though? Eventually, the data pool will be deep enough for us to come to a definitive conclusion.
On the flip side, it will be interesting — and probably depressing — to see the rim protection statistics for the Knicks’ other big men, namely Andrea Bargnani and Amar’e Stoudemire, in particular. Both Bargnani and Stoudemire are universally recognized as horrible defenders, especially given their size, but with SportVU, assuming they actually play, we will are almost certain to achieve clarity on exactly how bad they are. Granted, the rim protection metric won’t be able to account for the times when Stoudemire inexplicably runs the wrong way when defending a pick-and-roll — it usually feels like he is farther than five feet from the guy he is supposed to be defending — but the system should be able to help confirm what we already know.
Catch and Shoot
Catch and shoot stats measure how effective a player is at converting jump shot attempts from outside ten feet of the basket after no dribbles within two seconds of catching the ball. Notably, this metric differs from the “spot-up” category on Synergy. SportVU will allow us to see exactly how someone like J.R. Smith — a player whose abilities and usefulness are incessantly debated — does in catch-and-shoot versus isolation or post-up, so that we can discern something about how he should be adjusting his offensive approach. MOTE tells me that Smith is much better in catch-and-shoot as opposed to isolation and/or contested step-back jumpers, and though Synergy backs up that assertion to some degree, SportVU will give us even more ammunition when yelling at the television.
Pull-up jumper efficiency is extremely critical for the Knicks. The aforementioned Smith, Carmelo Anthony, Iman Shumpert and Tim Hardaway, Jr. all rely on upon pull-ups, to some degree, as part of their offensive games. (SportVU defines a pull-up as shooting after taking at least one dribble.) For ’Melo, however, this metric becomes even more important because of his propensity to take pull-up threes in transition. Sure, it’s a real crowd-pleaser when he converts those kinds of attempts, but it is also painfully frustrating when he misses, mostly because those shots come at the expense of open teammates running the floor alongside him. There has been a lack of consensus on whether Anthony is actually good at these shots, too, but between Synergy’s transition numbers and SportVU’s pull-up metric, even this debate should be rendered moot once the data set expands over time.
With the arrival of SportVU, no longer must passing be quantified solely by assists. Assists are a flawed metric, anyway. For example, great passes made to wide-open shooters who miss are not credited as assists. Once again, SportVU to the rescue! “Assist opportunities per game” charts how many potential assists a player amasses in a contest. Utilizing the system’s data on in-game touches in conjunction with a conventional analytic like USG%, we are able to track how Raymond Felton stacks up against other point guards in creating opportunities for his teammates. We can take his touches-per-game and potential assists, and use those numbers to compare his effectiveness as a facilitator. Taking things even further via SportVU, we can track Felton’s efficiency on drives to the hoop, and make reasonable inferences about his overall usefulness.
Taking a page out of the NHL’s book with respect to passing, SportVU also tracks ”secondary assists-per-game”. Defined as the “Quantity of passes made by a player to the player who earned an assist on a made shot. [Primary] assister must make a pass within 2 seconds and 1 dribble for the [original] passer to earn a secondary assist.” One only need be a casual Knicks’ fan to see the value of this metric. Take Anthony again, he of the reputation as an isolation-fiend. Had we been able to access SportVU during previous seasons, it would have been fascinating to see how ‘Melo graded out. Last season, in particular, the Knicks’ spread offense was so effective because of Carmelo’s ability to draw and recognize double teams in the post, and initiate perimeter ball movement by kicking it back out. Essentially, while Anthony’s assist numbers were nothing to write home about in 2012-13, he probably compiled secondary assists galore. And sure, Anthony wouldn’t have been credited when more than one pass was made after he passed out of double teams, but we surely would have a better perspective on just how valuable his passing was to the team’s overall offensive strategy.
The NBA’s analytics movement is clearly here to stay. Between the front offices, the media and the fans, we all have the opportunity to look at the game of basketball in ways that were impossible even 20 years ago. And as the data becomes more voluminous, we will be able to come to even better conclusions about the best ways to take advantage of what players do well and avoid what they don’t. Make no mistake, this is a big deal. SportVU is the next step in the evolution revolution, and we’re lucky to have access to such data, even if it’s in a limited fashion.
So jump on in, the data pool’s fine!