Digital & Dirt
00:31:37 · Ian Dallimore
[00:31:37] S0: I'd love to be inside of your brain. It's probably just like, let's go. Just constantly. It... But but, I mean, again, going back to it and and asking, like, a more detailed question. You know, we as we as fans now, we expect data. But what is it against staying with with tennis? You know, fans are so interested in things like ball speed, serve quality you had mentioned. So I'd love you to kind of elaborate that. And then even match momentum. Like, I think about the sports that I don't understand, and then you watch the momentum change. I think about watching World Cup and watching FIFA, like watching the changing of tides and watching what happens. Like, helps you... Your mind to understand a sport that maybe some people aren't familiar with. So then you become a passive fan. And then because of the data that you guys are providing, like at the US Open, you then start to understand it and then you embrace it and then you become a fan, which is why probably you guys have that partnership or they have that partnership with you guys. But what is it like serve quality as data point?
00:32:36 · Monica Ellingson
[00:32:36] S1: Well, serve quality is a brand new stat that we created this year. And, you know, the serve is definitely the most important stroke in tennis and there are so many different physical components that go into it. So, you know, it's the positioning of your feet, the loading as you... It's almost like a spring that, you know, loads and then springs forward, you know, trying to get that momentum, the energy from your lower body moved up all the way through the wrist snap. And so it's a difficult stroke to understand because of all these physical components. You know, I've... I have a a coach that I've worked with for a really long time for my tennis game, and I'm not a very good server. I don't have the patience to get out there and practice and what have you, but it's like, okay. You know, move your feet this way. Okay. Now now spring into the into the court. Now snap your wrist this way. And it... Just all these different pieces. It's a it's a difficult stroke to, you know, to be able to master. So to better understand it, we... For the first time, we gathered biometric data. And so if you look around the stadium, there are cameras that are typically, these are used for serve speed as well as the automated line calling, which they've had at the US Open since the pandemic times. And we're using those same same cameras from Hawkeye to gather biometric data. And so we gathered that biometric data, and then we used IBM Bob, which is one of our AI development tools and created a model for the different serve... The criteria that go into a serve. So created the criteria for the serve and then, you know, married that criteria for the actual effectiveness of a serve with the accuracy of the serve. And that came up into a serve quality score. But still as a fan, you're looking at that and you say, oh, serve quality 73%. What does that even mean? So we then use GenAI to create our natural language processing to be able to create a summary of the different criteria that went into that serve score. So what are the what are the physical components about that player serve that contributed to their serve quality? And that was what we bubbled up into the app. So it was a simple score with a couple of sentences describing why.
00:35:03 · Monica Ellingson, Ian Dallimore
And, you know, it was key that we kept it simple that way, you know, because it's, you know, not every fan is gonna gonna wanna dig deeper into that. But, you know, we just started this. I mean, literally in June, I saw a proof of concept for serve quality. We created a three d model of the court, the player, and the ball motion as well as part of this. And I said, hey. Well, you know, people have been doing this for kind of a while now in terms of, you know, visualizing the court, the movement, the motion on the court. But what can we actually... What insights can we pull from that? And within a matter of weeks, we had the serve quality score. We had the overall fan experience. We were able to test that and get... Enroll that into the digital platforms, which we released in August. So that's an extremely fast timeline timeline that we were able to go from proof of concept or idea all the way to launching the product. [00:36:04] S0: Yeah. Definitely shout out to your team on that because that is a... That could have been an easy like, okay, let's let's continue to test this beta internally and then let's launch it. But, yeah, that's that's fascinating. Do you ever do you ever try to use the same mechanics? You probably don't have the cameras and whatnot, but, like, on your own serve to be like, hey, coach. I got this.
00:36:04 · Ian Dallimore, Monica Ellingson
[00:36:04] S0: Yeah. Definitely shout out to your team on that because that is a... That could have been an easy like, okay, let's let's continue to test this beta internally and then let's launch it. But, yeah, that's that's fascinating. Do you ever do you ever try to use the same mechanics? You probably don't have the cameras and whatnot, but, like, on your own serve to be like, hey, coach. I got this. [00:36:24] S1: Not not yet. It exists. I mean, some coaches have that. You know, there are there are apps where you can, you know, hang your phone from the the fence and, you know, it will capture video and do some analytics on that. I personally haven't because I'm like, I know what I need to do. I just need to get out there and practice. [00:36:40] S0: What level are you by the way? [00:36:42] S1: I'm a USTA four o player. Mean, just, you know, just intermediate. [00:36:48] S0: I won't even bother you with the pickleball question because that's always [00:36:51] S1: Oh, I mean, pickleball is great. Anything to get people off their sofas and out there being active. [00:36:56] S0: But I have some diehard tennis friends that are like four fives, five o's, and Uh-huh. And you would think that they're like a bunch of hooligans showing up at the country club. It's just this... Like, everybody chill out. [00:37:07] S1: It's just
00:37:07 · Monica Ellingson, Ian Dallimore
[00:37:07] S1: It's just [00:37:07] S0: a sport. Alright. So IBM likelihood to win. This one is like... It's one of my favorite things that happen not just in tennis and and golf, but watching... I'm a big college basketball Duke fan. And I love when the likelihood to win, it's like 982%, and then it starts to shift. And then you see, to your point, the data points that you guys have, you see the momentum shifts in a game. And then so watching... Like, I I actually root for the likelihood to win, to lose and be wrong. But again, it's another data point that you guys... And I'd love for you to talk a little bit deeper about it, where I'm assuming you're using like historical data and and when these things occur and when this much time's left, but it's probably deeper than that. Like, is the first string in, the second string in, who's all these things. So... But the likelihood to win is one of my favorite data points that I know you guys use on a few things, and it's fun to watch it. Not be wrong, but sway back and forth because that's when it gets super exciting.
00:38:06 · Monica Ellingson
[00:38:06] S1: Yeah. I think the likelihood of when we've had for a few years now and and really it... You know, fans expect this. You know, they... They're looking for predictive, you know, performance measurements. And, you know, of course, everybody wants to see like, oh, IBM was wrong this time. But the thing that I really like that we we did this year was we actually two years ago, we changed the likelihood to win to be live. So to produce the likelihood to win, we are analyzing. At the US Open, we have access to twenty years of historical data. Of course, we have live stats that are coming in. And so these are updated after every single match. And then we're also processing what everyone in the industry. So what are the pundits talking about these particular players? And all that comes together to make the likelihood to win. And then after every single point in the match, we're updating that. So you can see this ebb and flow of momentum, which you mentioned earlier and which really helps with the storytelling. But we rolled out the live likelihood to win, which updated after every point two years ago. And I was looking at it and I said, but if you're a fan and let's say you stepped away for a few minutes and you come back in and you're like, what happened right here? What happened at that momentum shift? So we then we added in, you know, key... The key points, which basically talks about, at that particular momentum shift, what happened so that you can catch up and get a better understanding as the tug of war between two players occurs, you know, what happened at that particular point? What was the key that pushed it over the edge, you know, back over to, you know, the other player, etcetera? So we're just continuing to evolve that as a storytelling measure, you know, because like you said, I mean, every single sport has some level of momentum involved in it. And, you know, that's really these, you know, when that shift, that momentum shift occurs is think about like, you know, any kind of story you're trying to tell, you have a, you know, the buildup and then you have, you know, the climax. So it's the climax of the story. You know, what happened at that point, you know, and continue to evolve that and, you know, be able to help with these storytelling measures.