I HAVE A tendency to speak very fast when I'm nervous. So as I entered the Sonics' meeting room for my first draft meeting in late May 2008, I reminded myself to take deep breaths and stay calm. GM Sam Presti had hired me as his first analytics consultant, and I had come to Seattle to deliver my analysis. Seated around a table with Presti and his full basketball staff, I didn't want to sound like a geek on speed -- or, more accurately, a geek on speed who'd been cut from his eighth-grade basketball team.
At that time, maybe four or five NBA front offices used any kind of advanced statistical information, and even fewer had the analysts present at their pre-draft meetings. I'd been teaching at Menlo College in California with a Ph.D. in economics, and basketball, up until now, was more of a hobby. This was both my and the team's first real foray into live analytics-based draft discussions. I doubt anybody really knew how it would work. I had presented statistical analyses many times before, but usually at conferences to other economists -- people who understood the language of multiple linear regression and general equilibrium models. But this audience spoke basketball, a language I still had a long way to go in learning.
When it came my turn to speak, I explained why my analysis ranked Kevin Love as one of the top four players in the draft (he was a tremendous rebounder and shooter) and that Brook Lopez would be a total bust. (Hey, you can't win 'em all.) I argued that the data indicated that Russell Westbrook, even in limited minutes at point guard, had demonstrated the passing skills necessary to make the full-time switch from shooting guard. Overall the group seemed curious, even if it didn't totally buy into what I was saying.
After the meeting, I flew back home to San Francisco -- I worked for the team remotely -- but returned to town for draft day. That first year, the Sonics' final in Seattle, the team gave me a desk, a computer and a TV. My desk was not far from the draft room, but its door was closed. I watched the draft on TV. When the door did finally open, though, Presti came out with a big smile on his face, shook my hand and said, "Congratulations. You got your guy." I had been hammering the group for the previous two weeks with emails lobbying for Westbrook. I'll never know whether those arguments made a difference, but it was thrilling to be on the same side as the decision that Presti made.
This was the beginning of a seven-year run of consulting with NBA teams on how to build analytics into their decision-making. There has been tremendous progress during that time, as the gap between the people who speak analytics and the people who speak basketball has shrunk dramatically. Our languages have been merging. (Have you noticed how frequently TV commentators now use the word "efficient" during broadcasts?) But now, we've reached a crucial turning point. Today, teams are overwhelmed by the avalanche of data produced by SportVU cameras -- palm-sized devices that hang above the court in all 29 NBA arenas, constantly capturing the position of everything that moves below -- and they have little idea what to do with it. Most teams don't even have any idea how to hire the right people who would know what to do with it.
To understand how we got here, it's important to understand how analytics have grown in the league. After the 2008 draft, I spent the next season providing more and more analysis for the Thunder front office and the coaching staff. A year later, I returned to the draft meetings, which were now in Oklahoma City. In just that one year, the number of analysts employed by NBA teams had grown significantly -- there were probably 10 teams that were now paying attention. I spent the four days leading up to the draft presenting my analysis, discussing the pros and cons of each player in extreme detail, weighing in on trade scenarios and waiting around for something to happen.
The hours of waiting were often filled with watching film of prospects. It helped me refine my analysis, as I soaked up details from scouts that I never would have seen on my own. ("Rewind that. ... Did you see his foot placement there, getting ready for the rebound? That's NBA ready.") During one of these sessions, we were watching film of Syracuse point guard Jonny Flynn. I mentioned that, based on the rate at which he collected steals, he was likely a good defender. But one of the scouts explained that Flynn's steal total was likely higher than other point guards' because Syracuse played mostly zone defense, which allowed guards to attack the ball more. I checked that insight against the data and it seemed true, so I adjusted my defensive statistics to account for the dominant style of defense used by a player's team.
There was another difference this season. Now there was a seat for me at the table in the draft room. Over the next several seasons, more teams hired more analysts, and the data continued to improve. More and more of the concepts of analytics (such as scoring efficiency) seeped into mainstream dialogue. A copy of Dean Oliver's "Basketball on Paper" (the publication of which in 2004 really marks the birth of basketball analytics) could be found in just about every NBA front office. Now, not all of the copies had been read, but at least they were there.
After our trip to the NBA Finals in 2012, the Thunder decided they needed their analytics leader in Oklahoma City full time, which was certainly a positive for the field. They didn't, however, want to pay anywhere near what I would be willing to accept to move my family from San Francisco to OKC. They had already successfully recruited my former intern, a Stanford student, to accept a 60-plus-hour-a-week job for a salary that was far below what he could have easily gotten at a big consulting firm or investment bank, so they believed they could do the same with my position. At the time, they were probably right. There were still skilled people willing to take salaries well below their market value to work in the NBA. I left with no hard feelings.
So, after five seasons with the Sonics and Thunder, I hooked on with the Cavaliers. The process of being trusted was much quicker in Cleveland. Several members of the team's front office were actually eager to hear what the analytics said.
But soon a new challenge emerged: Starting in the 2013-14 season, SportVU cameras were installed in every NBA arena. The amount of data available to teams suddenly grew from a pond to an ocean. Think about it: Those cameras capture the coordinates of 10 players plus the ball 25 times every second. That's a vast amount of data. As a result, the race to unlock the secrets of the SportVU requires a much higher level of skill than what was needed when I first started working for the Sonics. In the beginning, anyone with advanced spreadsheet skills could probably add value to a front office. Now, though, deep statistical programming skills, along with advanced computer science knowledge, are needed to create value. These are skills for which companies such as Google and Facebook pay quite handsomely.
But teams have been slow to recognize the sandbox they're now playing in. The analytics community worked hard to be accepted and learn the language of the NBA, but maybe we did too good a job fitting in. Entry-level analysts are viewed not much differently from entry-level video assistants: lucky to be in the NBA and worth a salary not much higher than $35,000. Teams are used to competing with high school and college athletic departments for staff, not with McKinsey and Bain. Realistically, aspiring NBA analysts must be willing to take at least a 50 percent pay cut from what they could earn elsewhere.
Take Nick Martineau, one of my interns while I was with the Cavs. He was a perfect candidate for a full-time position we were trying to fill. He had played Division I basketball at BYU, had great people skills and was earning a master's degree in statistics. He had done some significant work on the SportVU data for us, and our coaching staff loved him. While I was pushing him forward, he got an offer from a venture capital firm that was double what the Cavs were planning to offer him. Instead of understanding how highly valued Nick's talents were, Cleveland took the stance that Nick just didn't want to work in the NBA enough and so the team let him walk. Nick's story is hardly unique. Whenever I've discussed stories like his with friends and colleagues across the league, I hear more of the same thing. It's not easy to turn down six figures for the privilege of working in the NBA. Teams spend top dollar on the very best head coaches and general managers, spend millions on practice facilities -- even the last guy off the bench -- but when it comes to analytics, there is no commitment to hiring the best of the best.
Of course, another challenge in hiring is the volatility inherent in any NBA job. I moved to ESPN after two years with the Cavs in large part for the peace of mind of having a stable job and the knowledge that my work wouldn't be suddenly interrupted by a change in regimes.
For all the challenges, though, there's no doubt that analytics are a big part of the NBA's future, especially as new ownership groups, often from the tech, hedge fund and venture capital worlds, enter the league. Some teams are starting to figure it out -- some even are raising salaries -- and departments are growing in number. And thanks to new streams of data, there are whole new languages to learn -- the biggest advantage will go to teams that want to learn them.