AI in Sports is a Game-Changer for Injury Prevention and Player Tracking

What's going on, everyone? It's Daniel Littwin, the voice of b to b right here live from Washington University in Saint Louis. I'm covering a sports business summit here today, and speaking with some of the presenters about their presentations and the major trends shaping the sports us world. Today, I'm joined by Ali Bray. She's head of product at Gemini Sports Analytics. She's gonna be giving a presentation later today called Money Ball in a bot using AI cloud applications for sports analytics. Obviously, very timely, so much potential there, and we're gonna get a pulse Allie today. Allie, how you doing today? Good. How are you? I'm great. Thank you so much for joining us. I'm really looking forward to learning from you later today, at your present Now, AI and cloud applications for sports analytics, is obviously the future. It's than now. It is a foundational aspect of any sort of intelligent decision making, informing everything from literal plays. To organizational and operational decision making. But I wanna hear from you what you see as some of the most in important layers where AI cloud applications are elevating the Game of Sports. Well, first of all, that was a great description Do you want my job? Oh, yeah. That's trade. Yeah. We could trade. But generally, sports is a really interesting field because obviously AI is all over society now. Right? And but in order to get to good AI decision making, you need to have really good data. So what some of the stuff we do at Gemini is how do we teams elevate their foundational data layer, their foundational data infrastructure so that it can then use that data to make good decisions off of it. So it's a big, like, there's a lot intertwined there between the, like, data and the foundation of what you're doing and then getting into the predictive analytics on top of that. I would say the biggest space for predictive analytics right now is in player acquisition, player evaluation, really anything around players. Because there's a lot of data there. And it's a little bit more black and white than what's happening on the field. So while there are lots of really good applications for player performance, team performance, tactics, what have you? We've seen that, like, on the recruitment, the scouting, the salary side, is definitely the first place teams are going right now. Interesting. Are you seeing that layer of AI supported data insights, starting now earlier in these players' processes or careers, right? Like, I'm talking high school, right? Like, our our plays or footage or, etcetera, being, analyzed through these lenses earlier in player's careers. Somewhat, Gemini doesn't deal with that directly. Sure. But I've worked at a couple other sports tech companies do film at the high school at the college level. We work with a couple college teams at Gemini. We're trying to use AI to enhance players in the transfer portal or how to make get the most out of their team and what, like, what strategy that they should use in a particular game type of thing? So we work more on the team level than the individual player level, but there are plenty of companies in the space that are at the player level and trying to use AI as, like, coaches even down to the youth level. It's something that I don't have a ton of personal experience in, but is huge in the industry. Yeah. Oh, I'm curious then what the launch of those kinds of data layers look like. Right? Are teams themselves sort of integrating and building out their own in house data teams to make this kind of analysis and informed decision making, or are they working with third party partners like yourselves to do it? Like, where is the energy, operationally right now? That's a really good question. And it honestly depends. Yeah. So every sport approaches it a little bit differently because every sport is at different stages in their data journey. Pretty much every team these days has some sort of a data department that can mean data engineers, data scientists, analysts, what have you. Baseball unsurprisingly because Money Ball came so early is the furthest along. So a lot of baseball teams have thirty, forty, fifty, person departments, which starts to mimic big tech companies. Mhmm. Every other sport has maybe one to three people. So they're way further way for, like, way further behind. They can't do as much. And then there are a bunch of companies that are trying to be that analysis source for teams, kind of like consultants. We sit in the middle in that we're trying to help teams in house data teams become better, faster, stronger, and reduce the time between question and answer from nontechnical stakeholders or people that are actually making decisions. So we're trying to supercharge the people and help them with the process and technology to become more mature data operations. Sure. Okay. Yeah. That makes sense. I wanna pick your brain a little bit more about the technology in action. Right? So advanced AI models now, especially like in basketball, for example, And I was picking my colleague's brain on this. He was giving me some more examples to work off of, but advanced AI models right are now helping through sort of tracking and computer modeling, helping track patterns and movement in plays, getting more insights into how teams, perform against other team types, other play types. Right? And this can really be, especially in basketball, delineated into sort of buckets of play types. Right? Helps give insights, helps coaching staff know if, you know, Bob Jones, the basketball player plays a, you know, does a pick and roll well or not. Should he keep doing it on the court? How well are we seeing AI models track some more unique unconventional, innovative plays in sports today? Are they smart enough to sort of think outside of the box yet, or are they giving more of a foundational layer and still in their nascent stages. Yeah. That's a really good question. I think, again, I I hate to keep saying it depends, but it kind of depends. I mean, that's fair. A big thing about AI generally and the way Gemini approaches is it approaches it is we like to say we are a combination of people and machines to make all of that better and help you make better decisions. So there are teams with really, really good data scientists at their club that are doing some really unique, really interesting things. We have some of our in house data scientists who are doing really, really neat, like, side, side work kind of things. That is different and not as foundational as kind of basic models. So it's out there and it's available. I I would say in force, there's a lot of things that you can do before you get into those, like, really niche areas that are potentially more impactful to the team. Yeah. So for in one of the things we recently do, and I'll touch on this in the presentation later. As we were working with the team in Europe who wanted to get promoted to the next league, and they wanted to understand how teams in the higher league played. And what different playing styles were, so they could try to mimic those playing styles to try to get promoted to the next league. Yeah. And then to go even deeper than that is once you understand you're playing style? What players fit into that playing style? And how can you predict how a player comes to your team and impacts that playing style? So, like, that's a very common pattern to go a couple layers deeper, off the original question you're asking to try to drive an impact and an outcome. Totally. Now what about, refining some of those data models and actually training them? Mhmm. Obviously per sport what you feed these AI models is going to be different. It's gonna be nuanced. The datasets and sort of the the path to a smart, you know, AI tool is long and nuanced. Give us a pulse check on what that looks like today from sport to sport. Right? How that nuance is key in building strong AI applications and Sykes. Absolutely. So that's kind of the crux of what we're doing at Gemini to in order to get good make good data driven decisions, you need to first have good data. And then Yeah. Yeah. It's not enough to just have data. And I think the sports industry has done really well in that there are tons of companies that are providing data across a large swath of different types of data. So event data, what's happening on the quarter pitch, gym data, performance data, so how fast a player's running, how much they're jumping, etcetera. Tracking data where they are, I can go on. Yeah. But what becomes hard is how do you align all of that data? How do you get it all talking to each other? So you know that Allie in this one data set is the same as Ally in this other data set. And Gemini is doing that for our teams, because generally that's a hard problem, because I might be Allie Brabant in one, Allison Brabant in another, Ali m Brabant in a third, and there's it's pretty hard to align those, and it's time consuming, and also frankly kind of boring. True. So we're solving that four teams and creating these foundational layers of data by sport. So we're part we have data partners with most of the data major data providers across all of the major sports, and our teams get that that aligned data and then some derived data on top of that for free. And then they can go into our platform and choose what types of data they want, massage it a little bit using their subject matter experts in house to say, I want to filter for the x, y, and z. Right. And these these different parameters, because that's what I care about looking at, and they can do that in a matter of minutes. Because they don't have to do any of the other coding work required to get the data in a clean aligned arranged spot for them. Right. And then they can go into the application, which is we have a web application that lets them run define what type of model they wanna run. And then we use AutoML under the hood. So for those that don't know, AutoML lets you run thirty to forty different models at once and ranks all of those models. So you're getting the best model that's currently available instead of having to do it by hand where you're manually coding one maybe two if you have a lot of time. Yeah. And just kind of hoping for the best that you chose the right one for this problem. And those thirty to forty models that we're running are automatically refresh daily. So we have the most up to date models in all of the common packages. So you're getting the best of kind of the industry at your fingertips. So you can run a model in twenty minutes. Whereas coding it by hand may take you weeks months if you're trying to align all the data as well. So you can iterate really quickly, and that's where the human comes back into the process and technology side of data maturity. Because then the human can go Oh, this doesn't look right. Let me try pulling out a feature, putting a feature back in because I think this should happen, and I can gut check this and iterate on it. Yeah. But you can do that in a matter of hours as opposed to a matter of weeks to months. Yeah. So you get to decisions and good data a lot faster. That's really cool to hear that the sort of operational side of data maturity is already at that stage. And, you know, operational teams are able to leverage that kind of insight. And, you know, have such confident data hygiene too is really cool. I wanna ask you one last question that it's a little bit more about the impact of data analytic and sort of, AI supported data insights on the actual feel of the game itself. Right? So I'm gonna use as a test case and, you know, don't kill the messenger here. I got this from my colleagues here, but James Harden's career. Sure. Let's talk James Harden. Okay. I feel like he is a good test case for min maxing basketball. And being successful at it, right? Through, and his kind of whole career is, premised around you know, he was identified as being really damn good at a few key plays. He was brought on team for that reason. Mhmm. He was successful at those plays. However, a lot of people, fans mostly, consumers of the sport complain about Well, he only does two plays ever, and he's boring to watch. So I'm curious what your thoughts are on sort of You don't have to really weigh in on that specifically, but just kind of like the the larger ecosystem of how AI actually impacts the game experience, right, how data insights and database decision making and sort of minmaxing, points on the board. How that impacts the sports experience? What are your thoughts on, maybe what's on the horizon for that? Challenges, opportunities? I don't know. What what comes to mind when I bring all of that up? The first thing that comes to mind is two things that are other examples that I would say are counter examples. You hit me with them. Yes. First is I'm a big baseball fan. K. And the change year in shot in pitch clocks and shortening the game was a very data driven decision, and people were very skeptical going into the season. And I think it's the best thing that happened to baseball. And I think from reading about it, fans largely agree. It made the game faster, more enjoyable. Like, more people are going like, it's I'm going to baseball games now. And I agree. The shot clock has been key. So yeah. Exactly. And then, like, another one is I'm a big Golden State Warriors fan, so Steph Curry is my guide. And you look at the increase in three pointers after Steph Curry, and that's, like, a data driven approach in basketball. You see, like, the mid range shot doesn't go that well. Most of the time. So, like, the two real outcomes are go to the net or hit a three. And that's all data driven. I think anytime you use data there could be potentially bad outcomes or outcomes people don't like. But I don't really that doesn't feel like a reason not to use data. In my opinion, granted I'm a bit of a data nerd, so take that with a grain of salt. But I think there are always applications. Every time sports change, people get a little bit upset about that because you're used to the way things work. But I think there's so many applications, especially for fans for their teams to get better or do more with the resources that they have. In ways that are really exciting. And I think there are potential downsides to that. Like, there always are. Sure. You can look at the shift in baseball, like, That was data driven, and it was not great for the game, in my opinion, because it made it kinda boring. And then you can go back on it. I think, like, the r of data throughout his history or throughout the industry, is it changes as you see the implications of it? And in a lot of ways, some of it's an experiment, and you're trying to see how how it works, and then you reevaluate, you come back, and you keep going. Yeah. And that kind of just feels life like. So I'm really excited to see where the industry goes. I think we're Still at the beginning in a lot of ways. Totally. Hey, fans out there. Maybe don't complain if your players start hitting more threes because of data. Right? I love it. Alright. Ali Bray. Thank you so much for your perspectives today. I really appreciate it. And I'm looking forward to hearing All of this contextualizing your presentation. If folks want to get in contact with you, learn a little bit more about the cutting edge of data and its impact on the sports experience, sports decision make How can they get into a truck and they learn more? Yeah. Absolutely. I mean, first, first of all, feel free to look me up, add me, message me on LinkedIn, Ali Brabant, also look up Gemini We're pretty young as a company, but doing some really cool things and have a really strong team. So Gemini Sports Analytics, and then feel free to shoot me an email as well. I'm always happy to talk to people trying to get into the industry. I had a bit of a roundabout path to both Data Tech and Sports. Cool. So happy to chat with people. My email is allie at Gemini dot a I. Fantastic. Alright. Thank you again. Folks, we've been chatting with Ali Bray Vance, head of product, and Gemini Sports Analytics analytics. Good looking and presentation, and I'll see you up there. Yeah. Sounds good. Thanks so much.