In late summer 2014, the NBA hired Jason Rosenfeld as its first Director of Basketball Analytics. Previously, the Director of Analytics for the Charlotte Hornets, Rosenfeld was tabbed to head the new group within the league, under the auspices of the Basketball Operations department. While the league had no shortages of people doing similar work in related areas[1. There is a sizable business analytics presence within the league, as well as another group responsible for monitoring and tracking refereeing decisions. That department is in charge of things like the Final Two Minute reports.], this new unit was empowered to specifically study the on court product. A graduate of Harvard with a major in Statistics and basketball experience ranging from the work with the Bobcats/Hornets, student-managing the Harvard team to working with the Yao Ming’s Shanghai Sharks while still in college, Rosenfeld’s purview is to help ensure the league has the information it needs to make decisions such as the reported changes to the schedule, and possibly sheds some light about why Silver is reticent to embrace 1-16 playoff seeding, citing travel concerns. One suspects it is not merely intuition that multiple cross-country flights in a short time span should say the C Bags and Warriors meet in a first round playoff series. Rather, Rosenfeld and his team have already studied the issue. He spoke with me via phone late in the 2014/15 season about his areas of responsibility thus far, the differences between working with a team and for the league itself, and the kinds of skills required to do this sort of work. What is are some of the initial priorities in this new job? Things like the schedule or back-to-backs. Adam (Silver) has publicly spoken about theses things. We could just make a change but now we’re not just guessing. It’s not like we’ll go to a meeting with the owners and just say “hey we think the players are performing differently in these types of scenarios.” We’re actually digging into the numbers and can see when they play four in five nights or have two days off, or when they travel two time zones instead of three time zones. We can really really drill into that. My background is in statistics so I can read academic papers and journals, critique them, and then replicate the studies and do further studies. So I think schedule and back-to-backs are a great example. The D League is another fascinating one. Just like teams are doing a fine job of using it to test players coaches and staff, we at the league are using it to test rules. If we want to implement coach’s challenges in the NBA one day, perhaps, it’s a great opportunity for us to learn if it might be successful. But we don’t just implement it in the D League and then say “well, that was successful.” No, we are studying how long does it take? How successful are coaches [at using challenges]? Where are the problems? And go from there. Speaking of scheduling and health and wellness, from conversations with people at STATS, I’ve been told you can start to “see” fatigue in player movement, is that something you are looking at? ‘Ok, it’s his 3rd game in four nights, he’s starting to slow down’ that kind of thing? There’s an X% degradation in athletic output, and so on? That’s exactly the sort of thing we’re looking into but we’re going much beyond that. It’s not just “third game in four nights.” It’s was it a home or road game, was it home or road on the night before, and then traveling how many time zones? Are they an older team or a younger team? We have a lot of good data and if we really really dig in, we can take the analysis out to several levels. What you said is another form of a research question, but now that I’m here and I have help, we’re really trying take that analysis to another level but you framed it well. What else are you working on? The medical side is another good example. That’s that’s obviously related to schedule , how many games you’re playing and how many back-to-backs and four-in-five nights. But also what tests were the players are going through and what correlations there are between the results and an increased rate of injuries. With all the different data collected on the players, we try to get a better handle on what correlation there might be between different metrics and injuries. How involved are you in evaluating or investigating various proposals for draft or lottery reform, stuff like Mike Zarren’s “draft wheel” and stuff like that? That’s another example of something that we’re involved in. We will go through a lot of different options and proposals. We’ll run numbers and simulations, we’ll look to historically “ok if the lottery looked like this, where would the picks have gone in the 1990s and 2000s?” The lottery is a very good example of what we’ll spend time on to present a lot of different options for executive presentations to the board of governors or the Competition Committee.I think another big part of her role is when Competition Committee meetings occur or Board of Governor meetings, they have all the data at the disposal so they aren’t guessing. We’ll do a lot of preparation to arm them with the facts for topics whether it’s on the lottery or conference [realignment] or injuries whatever it may be, scheduling. You have a fair amount of lead time to prepare, it’s not like the day before the meeting someone comes in and says “hey, we’re talking about moving the three point line, what can you tell us?” We know when the important parts are on the calendar, we know the major topics also. With that said I haven’t even been through a full year cycle yet. And I think you might wanted to ask about this but it’s very different from [working with] a team where you know that there are these really important times of the year with the draft, free agency, and the trade deadline while at the league office, for me I’m still learning when what occurs and then where we could be most useful. Is that the biggest adjustment from the team side to the league? Adjusting to the different rhythm? It’s completely different. When your with a team you’re really it’s a lot of long-term research, getting ready for those those big decisions. Here, because I’m still new, it’s learning all the decisions, what they are, what we need to do to provide value. When you’re with a team you’re trying to figure out who to draft or who to trade or who to sign. I’m not really worried about player evaluation so much here, or for preparing for the game like of a lot of analytics guys do. So the nature of the questions are much different, I’m more concerned with the product of the league and doing analysis that will support decisions to improve the product. What are some of the skills necessary to work in the department? The main thing is technical skills both from a statistics standpoint and a coding. So statistical analysis skills are very important because if you’re looking at the scheduling questions you wanted to talk about how many time zones they’ve traveled through and how many days off you’re to start getting into multilevel regression, right? You get beyond this middle school math. So stats skills are for sure important but also coding because if you want to know how many days off they had at a time or how many time zones they traveled through, you probably don’t get the data in just the form you need it. You need to go through a lot of manipulations get it where you be able to analyze it. So that’s also important, obviously. But hopefully you really like basketball and understand the game, because if you don’t you could miss a lot of complexities in these problems. How much basketball to you watch now? I imagine when you worked with the then-Bobcats you watched all or most of those games, but how much are you able to watch now? Yeah, I went to all 41, I think if you can believe it? If you want to believe I went to every home game over two years, 82 games. Now I’m spreading the wealth around the league more. I still watch a ton of games but I’m watching more and more teams because you are so much less focused on specific group of 12 to 15 guys. You focus more on all the players or the rookies or the international guys or just the league as a whole. So when you’re with a team, I imagine you’re watching in a specific way, okay my numbers show this I want to see if I can pick that up. Now that you’re with the league are you doing the same thing? For example with the fatigue stuff, this team or player is in that danger zone, “can I pick that up with the naked eye?” Is that the sort of question you ask? I would say yes, but as you know somethings are just so hard to pick up with the naked eye. It’s like in baseball, a .300 hitter vs a .270 hitter, go from a great player to a terrible player, and you’d never know if you just watched them. Some things are really tricky like that. Some things are easier to pick up. If you’re talking about teams that play at a really high pace or just shoot threes or at the rim, that when you’re watching the game it might be obvious and you say “wow that is obvious.” And it backs up and is in line with what my numbers are saying. Something that gets lost a lot of the time is even people who are quote “analytics guys” and myself definitely included. We’re all pretty big basketball fans (emphasis his). Even if I weren’t working for the NBA, I’m sure I’d be watching a lot of basketball because it’s an interest of mine and something I like to do. That thing where people say people who do analytics don’t really know anything about basketball when in fact we’re watching 20 games a week? Sure, I love basketball. With that said my basketball knowledge and my ability to remember plays and games, just hearing someone like Kiki talk about that, I’ll never be at that level. I think it’s a real form of genius, I mean it’s really amazing. And I’ll never get to that level, but with that said, I like to think I’m a big fan in my own right and still watch a lot just out of interest and then hopefully maintain a pretty good understanding of the game. Working with a team and now the league and before that as a manager at Harvard do you feel your technical understanding of the game has improved? Not so much from an “analytical” standpoint just from understanding what’s going on out there? Has it been sort of like a graduate-level program in basketball. All the different experiences have helped me in different aspects. First at Harvard there was a terrific coach in Tommy Amaker. Second, by being a student manager you’re at every single practice. You really get to know what this team wants to do. From play calls to discussing strategy. You learn almost what it’s like to for the players, you’re around the team so much. So that was a terrific experience. Then when you’re working with [an NBA] team, you can get involved in strategy, how to prepare for games. you look at things a certain way. And then here with the league, you’re focused on rule changes, and how long things take, are they efficient or effective? And that’s all added to my knowledge. Again I would never put my knowledge on par with people like Kiki who have played and are real real basketball experts. But hopefully they will take my level of basketball knowledge and I’m going to continue to learn. It’s a lifelong process. You mentioned how long things take. Every year we hear about whether it’s replay or end of games or timeouts or anything to make it shorter, is that one of those things you’re involved with looking into? How long a replay takes? How much time does the D League’s “advance” rule save if we implemented it? That’s a great example. Because if you don’t have data you might say ‘okay if we added the “advance” to the NBA, that actually added 1 minute or two percent to game time’ or something. But we’re actually amassing data on every “advance” called in the D League this season. So we know on average they took X seconds, but they took Z seconds at the start of the year but only Y seconds in March or April, so we think the amount of it would take would go down over time. We’re collecting all that because you want to have expected value in terms of how much time with will save and as you know length of game and game flow has been one of the major topics out there.[3. Some of the rules tested during the WNBA “analytics scrimmage” between the Mystics and Lynx in May were explicitly about pace of play. I estimated the streamlined free throw rules tested would save about 8 to 12 minutes of running time from an average NBA game. There were clocks on the ends of benches and underneath the shot clock during Vegas Summer League, apparently for the purpose of malong timeouts ane other stoppages of more uniform length. No official comment as yet on where those or other time saving mechanisms might be along the process to adoption.] Is there any one thing you’ve learned in this job, something that’s working better than you thought it was from the outside for example? I think I would just say I’m very impressed overall. A lot of people really love the game, love the NBA. It just speaks to the NBA’s commitment to get better and better. If I weren’t here, if they didn’t hire me, the league will do fine. But the fact they want to really invest in analytics means they don’t want to just do ‘fine’. They aren’t complacent, they want to get better and better and use the power of data and analysis to get better results and maximize the probability of success like a good statistician would.