Chasing Perfection: A Behind-the-Scenes Look at the High-Stakes Game of Creating an NBA Champion

by Andy Glockner

  “Like zone recognition: I will concede we were terrible at that for a while, but we made it a much bigger emphasis and we’re doing much better on it now. It’s still a tough one, but we’re doing it.

  “With respect to the computer, the machine-type recognition, maybe it will give you better accuracy, and I suppose that’s inevitable given the way technology marches forward. I’ll concede that. And when that happens, I’ll be the first one standing in line and using that technology instead of the people, because of the second benefit: it’s cheaper.”

  That’s not to say that Synergy is shying away from technological innovation. During our conversation, Barr noted that Synergy was on the brink of introducing a more integrated product where NBA teams will be able to operate their own databases behind Synergy’s firewall. That will allow the teams to seamlessly integrate their work with Synergy’s data, along with Synergy’s video editing and other capabilities, instead of the team having to implement that kind of integration on their end. As part of that push, Synergy will start enabling SportVU data intake within their own products. Barr believes a total turnkey solution on that level will be really appealing to almost every one of Synergy’s existing customers.

  “A lot of NBA teams don’t want to build a database and hire programmers and do all of that themselves,” he said. “It’s very expensive and, ultimately, is inefficient. If you spend enough money on it, you get what Daryl [Morey in Houston] has. You get what Boston has. You get what Dallas has. There are some teams that are doing extraordinary things there. But I will bet that over 90 percent of the teams will end up operating their database, with their own proprietary tagging and services they buy into and import data, they’re all going to do it within Synergy within the next five years. Probably quicker than that once we have it, because the price point is going to be so much sweeter. You don’t need programmers to query the data. We allow methods where you can use an expression editor in order to create the queries without having any particular expertise. I think it’s going to be a game-changer.

  “We are not logging everything we could log [currently],” Barr added. “We’re going to identify several new things to make our data set more robust, but clearly there’s more stuff that can be identified. There is automated tracking like SportVU, which we are not currently working with in our system. So, yeah, if you have unlimited resources, and I think Daryl is the poster child for this, then you are going to be able to extract an nth degree of data, significant data and value into your applications and your ability ultimately to win games via the draft or scouting or trades or even your coach’s development. All of that stuff. But I do agree that in the vast majority of cases, teams find what we offer to be enough.”

  While Synergy was nailing down an alliance with the NBA in 2008, that also was the year STATS, LLC, began investigating the merits of SportVU, a company that was repurposing Israeli missile-tracking technology for sporting applications. STATS ultimately decided to acquire SportVU, which had adapted a system originally created by military defense electronics manufacturer Elbit Technologies, and was, in 2008, using it to track the movements of soccer players.

  The technology, which was created to analyze the actual paths of launched missiles against their anticipated trajectories, evolved to where cameras were hung above playing services to track the movements of players and the ball in three dimensions, and record all of that data in real time. When Brian Kopp, now the North American president of Catapult Sports, an Australia-based maker of wearable sports technology, arrived at STATS that year to work as a strategist, he was handed the task of figuring out how to translate this new soccer technology for other sports.

  “So we looked at American football, [and worked with] the NFL for a while, and early on a couple of basketball teams [came down]—Houston and the guys at San Antonio and a couple others,” Kopp said. “They reached out to me and took my calls because they were scouring what was going on in other sports, and they saw these camera technologies used in soccer, and they allowed me to put some cameras in the arena and start to go about building this technology.

  “What’s interesting about it is the technology didn’t exist for basketball. We had to build it, but it’s not like we had a basketball arena. So we had to go to the teams and say, ‘Can I please put some cameras in, so that I can build this thing?’ So when I first started [landing teams as clients], they didn’t have anything [in place to use it]. It was just an idea. Their edge was to be the first ones to understand it, to be the first ones to get access to it, and then hopefully drive it forward . . . they knew that eventually the power would be getting it in every arena, but they were hoping that [they’d] have a first-mover advantage, and so we worked with a couple of teams.

  “Our focus was the technology side, but early on I knew we had to build out the rest of the analytics off the back of it. It wasn’t just going to be, ‘OK, here’s a bunch of X, Y coordinates, and I’m just going to dump it on your desk, and [say] ‘go nuts . . .’ [although] that’s all [then Houston Rockets vice president] Sam Hinkie wanted . . . and teams like the Mavericks didn’t want me to work with anyone else.

  “I remember clearly having a conversation with Mark Cuban, and he said, ‘Well, I don’t want you to get too many more teams.’ And I said, ‘Well, if it’s OK with you, I’m going to try and get them all.’ For them, they were, ‘Wow, we don’t want it to grow too big,’ but to me, the advantage is never going to be an access to the data. It should be in what you do with it.”

  Kopp said that a lot of those early moments for SportVU were spent defining the basics of what was actually available from the data and what kinds of reports they could generate. Kopp believes that the creation of this more manageable information, combined with STATS management not supporting the development of more sophisticated data services off of its own data, is what led to the formation of companies like Second Spectrum and today’s whole industry of third-party data solution providers. Instead of expanding STATS’ own business, Kopp had to actively seek out external partners in order to get STATS’ data, in manageable formats, into the hands of potential customers.

  “To be very blunt with you . . . it’s part of the reason I’m not at STATS [anymore]. They didn’t support it from the beginning, so I had to go partner with other people,” Kopp said. “They didn’t allow me to build out my own Second Spectrum for a while, so I had to partner with people like that to broaden the aperture and exposure of the data. So I had to work with Kirk Goldsberry (a Harvard lecturer who also contributed to ESPN’s Grantland) to give him data and get it out there and have it exposed to people.

  “Ideally, you do that all in-house, you keep it all yourself. But I also knew that anybody who wanted to have access to this data would have to work through us. So, in a weird way, Second Spectrum, we gave them free data, they decide to—I like those guys, they’re really smart, [but] I really don’t like how they built their business, personally. Because they took data that I gave them for free, and they decided to run a business and I said, ‘OK, we can work out some [deal] to get access to it,’ and they decided, ‘No, we don’t need you anymore.’

  “Certainly, it could’ve been handled in different ways, both by STATS and by some of those potential partners,” Kopp added. “I could have kept it hidden in a vacuum, and people at Second Spectrum would argue, ‘Well, you wouldn’t have gotten certain clients without exposure to our data,’ and I’m like, ‘Nah, I’m pretty sure we would’ve gotten them anyway.’”

  (Asked to respond to this, Second Spectrum’s Maheswaran declined comment.)

  Early in the SportVU buildout, Kopp tried to sell the product directly to the NBA, but the league wasn’t interested at that point in widespread implementation. Instead, Kopp built an individual team client base from the couple of early adopters into fifteen of the league’s thirty teams. At that point, the NBA decided to license the product at the league level, putting cameras in all twenty-nine NBA arenas (the Los Angeles Lakers and Clipp
ers share Staples Center) and distributing the data to all of its franchises. The league, per an article by David Aldridge on NBA.com, actually started testing SportVU during the 2009 NBA Finals. In that series, the league used the technology to slow-motion check a crucial goaltending call, and it turned out that the referees had gotten the call correct. That sparked the realization at the league level of how powerful the technology could be.

  The NBA’s decision to install the cameras in every arena has pushed the league headfirst into the Big Data era, and much like it has with Synergy’s products, the NBA has now incorporated a ton of Sport-VU’s data into NBA.com, so fans (and media) can access it. The data portion of the league’s website is remarkably robust, with users able to check traditional statistics as well as a litany of advanced analytics options, plus huge video libraries, dynamic player-tracking graphics, and much more. While what’s accessible on NBA.com is just a fraction of the output generated, there really isn’t very much even the hardest of hard core fans are missing if they know how to navigate the site and the myriad SAP-powered search options for each category.

  Want to know the Bulls’ best and worst player combinations? How about which two Cavaliers paired the best with LeBron? You can find how the Grizzlies did when Marc Gasol was on and off the court. You can slice up Steph Curry’s 3-point shooting by court area. You can learn that Kobe Bryant shot just 32.9 percent on jump shots in 2014–15, and that DeAndre Jordan led the league in defensive rebound conversion rate. You can literally spend hours at a time combing through the various categories. The league also has a writer, John Schuhmann, specifically dedicated to crafting analytics-based stories for the site, to add even more refinement.

  Video availability is the gem of the revamped NBA.com site, though, especially with the disappearance of Synergy’s independent, public-facing product called MySynergySports. (In August 2015, Barr hinted that it may return in a different form down the road.) Now, on NBA.com, with the help of Synergy, you can pull up all sorts of clips to complement the stats you’re examining. For example, if you want to view, in sequence, all 202 baskets Curry made from five feet or closer to the rim in the 2014–15 season, you just need to click on the video option link for that particular stat, and they’re cued up for you. Overall, the site is an incredible trove of numerical and visual information that brings users as deep as they want to go into the sport, and this reflects the pro-technology reign of newish league commissioner Adam Silver.

  So, if fans now have this kind of access, just imagine what NBA teams actually have, and the many different ways they are using it in order to create and sustain competitive advantages.

  CHAPTER 3

  Analytics Believers and Doubters

  Teams are trending towards taking less mid-range shots, and that appears to be a sound strategy. So, these teams’ efficiency attacking the rim and knocking down perimeter shots should be a good indicator of the teams’ success (at least on offense). If restricted area and 3-point shots are the most efficient looks on offense, and teams are trending towards taking more of these shots, then a team’s ability to defend these areas must also increase in importance.

  —Dr. Stephen Shea, mathematician and coauthor of Basketball Analytics: Objective and Efficient Strategies for Understanding How Teams Win

  With every NBA team having equal access to the raw SportVU data and subscribing to some level of Synergy’s service as well, it appears that Catapult’s Brian Kopp was right: teams derive their advantages from what they can find in the data and how creatively they can implement ideas on and off the court. How they’re doing it, though, varies very widely from team to team, and a lot depends on the amount of money invested and proprietary work being done at the franchise.

  In Kevin Pelton’s February 2015 NBA analytics rankings, he labeled twelve of the league’s thirty teams as being either “believers” or “all-in” on analytics (which foots fairly closely to what research for this book determined independently). The four teams described as “all-in” were the Philadelphia 76ers, Houston Rockets, San Antonio Spurs, and Dallas Mavericks, all of whom also made the overall top ten in ESPN’s 122-team combined ranking of NBA, NFL, MLB, and NHL teams. Those four NBA teams also double as perhaps the league’s most secretive.

  The difficulty in establishing long-term strategic advantages, though, stems from the simple fact that, eventually, you have to show people what you’re doing (except in the areas of player health and wellness, which is why according to a growing number of NBA personnel, developments in that area ultimately will dwarf the current evolution of on-court strategy). Whether it’s in player acquisition, draft philosophy, or on-court strategy, there’s going to be a video and paper trail that everyone else in the league can dissect, and they can eventually figure out some part of your plan.

  That ability to dissect applies to writers, as well, so below are breakdowns of some unique ways in which three of those four all-in teams are thinking differently about conventional NBA strategies.

  The Spurs and (the One-Season Fall of) Strategic Roster Management

  The cramped interview area tucked just outside the visitor’s locker room at Denver’s Pepsi Center felt like the perfect setting for legendarily gruff San Antonio Spurs head coach Gregg Popovich to discuss his team’s highly compressed early-season schedule. It was mid-December, just six weeks into the 2014–15 season, but Popovich and the Spurs—who for years had been well ahead of the rest of the NBA in terms of managing their roster to endure the strains of an eighty-two-game regular season while still winning enough to position themselves for postseason success—seemed to be meeting their match in terms of balancing performance and player workloads.

  The Spurs entered this particular tilt with the homestanding Nuggets with a perfectly reasonable record of 16–7, which projected to around fifty-seven wins, which was a typically normal Spurs haul in the Popovich era. Popovich’s concern, though, was that the schedule, along with a spate of injuries, was already taxing his team—especially his aging core of Tim Duncan, Manu Ginobili, and Tony Parker—to a degree that was much worse than in previous seasons. November already had seen the Spurs play eleven games over an eighteen-day stretch, and this meeting with the Nuggets was in the midst of an absolutely punishing December run that, if you included a game on November 30 at Boston, featured nineteen games in thirty-two days, including seven sets of back-to-back contests. To make matters worse, none of those back-to-backs even featured two consecutive home games.

  “The whole month is just ridiculous. We just have to deal with it,” Popovich said, plainly.

  While coaches often can trend toward both hyperbole and recency effect, there was ample reason to believe Popovich when he said that December was the worst schedule month he had ever seen in his two decades with the franchise. In addition to the sheer number of games, the travel, and the quick turnarounds, a rash of injuries had left him much shorter on personnel than the ordinarily deep Spurs liked to find themselves. Parker, the starting point guard, was struggling with a hamstring problem that had cost him some games, and he wouldn’t play on this night. His primary backup, Patty Mills, still hadn’t at this point returned from summer rotator cuff surgery, so the point guard role was nominally in the hands of the inexperienced Cory Joseph, with some help from Ginobili and other perimeter players.

  The Spurs also had seen small forward Kawhi Leonard suffer a hand injury that cost him two games ahead of this particular contest (in which he played), and then soon after saw him miss fifteen more. San Antonio also had been without center Tiago Splitter for all but ten minutes of the season prior to his return on December 8, and on this night, he still hadn’t completely been integrated back into the rotation as he got his fitness levels back to normal playing standard.

  All of this meant Duncan, in particular, was finding himself playing more minutes than usual. Entering this particular game, Duncan had played at least thirty-six minutes in four of his last five appearances after not even averaging thirty minute
s a game during the 2013–14 season. Two nights before, Duncan played a then season-high forty minutes against the Los Angeles Lakers; later in the week, he would play forty-eight and forty-three minutes, respectively, in triple-overtime losses to Memphis and Portland. Popovich ultimately found a way to rest Duncan for four separate games in December, but the thirty-eight-year-old still played in fourteen contests and averaged 34.5 minutes a game for the month. It was hardly ideal.

  “It’s a little tougher on us this year, you know?” Popovich said, expounding on the impact of the schedule and the personnel issues. “Patty Mills is such a big part of what we do, coming off the bench if Tony didn’t play, because he scores. And Tiago’s just coming back into the feel of the game. He hadn’t played in, well, I don’t know how long now. So we’re doing OK. We’re much better off with Tiago and Patty Mills [though] if we sit people.”

  Sitting players in strategic fashion has been a major part of how Popovich has extended the contention window for this Spurs team as its core ages. The team even made some light of the strategy late in the 2011–12 season, which was a compressed sixty-six-game schedule after extended negotiations over a new collective bargaining agreement cost the league part of the campaign. Box scores list the reason players who didn’t enter a game received a “DNP” (did not play) in that particular contest, and against the 76ers on March 25, 2012, the Spurs listed Duncan’s reason for his DNP as “old.”