JEREMY: I think when most people think sports analytics, what they generally kind of think about is Moneyball. They start thinking about performance analytics, on how to use data to really make team and individual players better. And it feels like it was about 20 to 30 years ago when this really started becoming a thing. It started getting into both professional team psyches, but also fan psyches. Can you give us the high-level view of how that's evolved since then and kind of where we are now?

KELSEY: Yes, for sure. Moneyball, you really can't talk about sports analytics without mentioning Brad Pitt and Jonah Hill. Who knew they would motivate an entire generation of statisticians? But we'll talk about it briefly. So, what we do now is much more detailed than what they had to do when they were first trying to introduce sports analytics as a topic, or even as a field that people could get into. You have your Bill James and your Dean Olivers of the world. They were kind of the pioneers that really did the hard work in the sense of convincing front offices that they need to use analytics, and they need to use stats beyond just the box score or runs scored to actually analyze players, especially when it comes to comparing players and seeing who's more effective in different lineups and situations and things like that. I would consider that kind of the initial revolution where it was really game-changing for every single team to realize that they need to have someone, at least one person in the front office, whether it's on the coaching staff or someone that travels with the team, actually dedicated to data and stats, and to kind of spin that into what's going on now.

Every team, for the most part, has at least one person, like I said, but now people are getting so detailed to where they have someone who is focused on player health specifically, or you know, research and development and really specific fields. There’s still tons to learn with tracking data, we haven't really even scratched the surface in basketball in terms of how to really quantify defense and things like that. Tons of room to grow, but every change now will be much more minor compared to the initial kind of revolution of talking about points per possession instead of points per game, to really understand how someone is more impactful than another player. So yes, big changes will continue to happen, especially with new data sources. But that's kind of my spiel on Moneyball.

J: Turning away from the performance analytics, my feeling is it would surprise most people to know how much data science is used, especially at the professional level inside of organizations beyond just the athletic side and really more and more throughout the organization. Can you give an idea at the broad level of how it's being used, and why it feels more and more that this is becoming just a way you need to do business at that level?

K: Yeah, definitely. So obviously, I'm very pro-business side analytics. I think we have about, I would say 20 people, maybe 25, including interns and part-time employees, between our team, which is the ticket analytics and pricing side. Then we also have a really well built-out business intelligence side of things where they deal with a lot of similar data, but they're more focused on the customer journey and helping the sales reps, whereas we're a bit more revenue/ticketing side. Between our two teams we have 20 data scientists, some more data engineers, some more data analysts, BI analysts. But that's a pretty built-out team.

I don't think any front office in terms of sports has 20 people purely dedicated to data. And there could be a lot of reasons for that. It's easier to justify when we're directly

talking about revenue and when our boss's boss wants to know how much we made at a game, they need the tabular dashboards that our great team has built out, to be updated pretty much in real-time. So I think, and we'll talk more about the details,

but things like retention models, they are pretty common in a lot of industries, obviously. Like when we learned it at the Duke program, churn models are needed everywhere, but especially in our industry where we're talking about season ticket members. They have every single year the option to renew or the option to leave. We can very easily analyze that behavior. But I would say that almost every team has some version of a retention model, which is pretty amazing to think about because again, five, 10 years ago, that was groundbreaking stuff. So now we just get to kind of grow on that and evolve and model even more things.

J: I admittedly did not realize your team was that big. That's not just sports organizations, that's larger than a lot of very large Fortune 500 in terms of the core analytics team. That's incredible. So, I think you mentioned part of that is this world

of the pricing side of things, and this is one where sports, and entertainment more broadly, there's a very specific similarity to pricing airline tickets and pricing hotels. At least that's the feel in the sense that you have a set number of seats and you'd like those seats to be filled. So you're in a game where you're almost going for the “Goldilocks” price that you want it to be just right. You want it high enough so that you're making as much revenue as possible, but you don't want it too high such that you have empty seats. And my feel is, especially in your domain—where how the team does

throughout the year, who they're playing on any given night, whether they're playing on a weekend or a weekday— there's so much that goes into trying to figure out

what pricing looks like. Before we get into some of the more detailed pieces on how you think about pricing and use data, can you give the kind of high-level piece on how you would even start to start thinking about that question given so many things moving at once?

K: Yes, so because of how volatile the individual ticket market can be—we're talking about all these factors of what players are in, what's going on in the news,

all those things—that's why we really do want to rely on our season ticket member base. So, in most arenas and stadiums, you aim to have at least half, if not closer to three quarters, 75-80% of your stadium, arena, whatever it is, of your venue, already signed on before the season even starts. Those are your season ticket holders, whether they bought the whole season, half, some type of a plan. That really protects you

from the fear of one of your stars getting hurt or leaving or something like that. And that's why when you ask the “Goldilocks” question—which I like thinking of it in that way, in terms of maximizing revenue—obviously, we want to bring in as much revenue as we can. But especially for us having just moved, we were in New Jersey a decade ago and now we're kind of trying to rebuild a fan base here in Brooklyn, that's really tough because we need to have pricing that's fair to get the people that actually want to be in the arena here, while also keeping obviously ownership and people paying the bills, making money and having that revenue.

So, I can go into a quick example of how we try to find the “Goldilocks” price, knowing that it's very elusive and likely will never be exact, but we have sales data obviously, so we can see that your seat sold for a $100, two days before the game to let's say the Warriors. And then I can also see that someone posted a seat very similar to that, maybe right next to it for $200 on the secondary market, and it didn't sell at all. So, you could argue that the $100 could be too low of a price, right? Maybe they're willing to go $150, $180, but we know that $200 was too high. So that's a very, first of all, cheap example, I don't know that we had any tickets quite that low for the Warriors. That's an example of, at a very small scale, how you can try to find and what our whole pricing team's job is, just to try to find that in-between, sweet spot.

J: Where do you see sports analytics moving in the next 10 to 20 years?

K: Yes, hopefully still growing. Obviously, I think every new data source you get

will kind of change the whole landscape. So, I just finished Seth Partnow’s book, "The Mid-range Theory," and in that, he talks about how if we can start collecting audio on how players talk to each other throughout a game or coaches, then maybe we can get a better feel for if they're good at defense, if they're a good teammate. And it got me thinking about if we had audio in arenas, right? Like I know that they have the “clap-o-meter”, whatever tracking how loud the arena is, but maybe a more accurate thing to actually say is, this section is your most loyal fans because they are the loudest no matter what's going on in the game or different pieces like that. I always thought it would be interesting to have an indicator on your seats so that if you stand up, we know how much time you're actually spending sitting as opposed to maybe you jump up every time there's anything that happens, every dunk or maybe you leave to go eat food for a whole quarter.

Different things like that. So, maybe more in-arena tracking, I think would be really interesting outside of just buying behaviors. I know when we were talking earlier, we were talking a little bit about facial recognition. You could use that for a lot of things, but I think it'd be interesting to use it to see how excited a fan is or how engaged they are in different spots too. The more data we can collect, the more interesting our projects become. So selfishly, that's where I hope it goes.