Find out how data is revolutionizing sports to gain a competitive edge. Hosts Becca DeGregorio and Todd Whitney speak to Pierre d’Imbleval and Halee Mason.
In sports, the competition is fierce. But athletes and their coaches have found new ways to aim higher, achieve more, and respond to the rapidly evolving competitive landscape. How? With valuable insights from data specific to their teams, players, and strategies. In this episode, we’ll dive into the new approaches to data that are revolutionizing gameplay featuring interviews with Pierre d'Imbleval, chief information officer with Renault Sport Racing, and Halee Mason, lead data scientist for Cloud9.
For more information about Microsoft's In Culture podcast, please visit: microsoft.com/inculture/podcast
Todd Whitney: Today's episode is very broadly about sports. Becca, weren't you training for something not too long ago?
Becca DeGregorio: Yeah, a couple of years ago I trained for my first marathon.
Todd Whitney: Okay, that's cool. I've never run a marathon. I'm probably never going to run a marathon, but I'm kind of curious about… how do you do it? How do you prepare? How do you train to run one?
Becca DeGregorio: Well, it helps to run with someone. I ran with a friend, and we just kind of took it day by day, lots of running.
Todd Whitney: Naturally.
Becca DeGregorio: Cross training, drinking tons of water.
Todd Whitney: Hydration.
Becca DeGregorio: Carbo loading. We researched the course and we bought weird running accessories to sort of do all of this in, but all of this is to say that we really DIY'd the thing and it was kind of a crap shoot on race day. We weren't sure if we were going to finish it and there were definitely thousands of people running ahead of us that took it way more seriously than we did.
Todd Whitney: What do you think they did in their training that you didn't do?
Becca DeGregorio: Well, they probably didn't skip their runs on Saturday mornings like we did. I hear consistency is key. Getting up the same time every day, eating the same things, wearing the same clothes, yada, yada, yada.
Todd Whitney: Okay, I hear you. Those are all great things, but what if everybody's doing them? Consistency is there across the leaderboard, but what next?
Becca DeGregorio: Well, I don't know. I guess you just train harder.
Todd Whitney: But those runners are already at the top of their game. They can't go any harder.
Becca DeGregorio: Yeah, I guess there's only so much running and carbo loading and hydrating you can do that everyone else is matching you on. Maybe train smarter, but what does that really even mean?
Todd Whitney: Yeah, you're getting there. When a sport is your profession, you kind of hit a performance ceiling with all the stuff that you can do on your own. That's kind of like one of those conundrums that happens in all types of sports. You get so good at dribbling a basketball or passing a soccer ball or racing a car or playing a video game, and then the edge gets thinner and thinner against competition when everyone's doing the same thing. Exploring new dimensions of training takes innovation and creativity, and one of those dimensions is data.
Becca DeGregorio: I see where you're going with this.
Todd Whitney: Mm-hmm. On this episode, we'll hear about how two very different sports are approaching data, how it's feeding into strategy, how it's helping coaches and athletes gain new insights from competition, and ultimately how it's helping athletes win. We'll speak to the chief information officer of a Formula 1 team and a data scientist pushing things forward at an e-sports organization. For Microsoft, this is In Culture. I'm Todd Whitney.
Becca DeGregorio: I'm Becca DeGregorio. You don't have to know a ton about Formula One to appreciate the role that technology plays in it. Racing cars is a marvel of engineering, but reaching the top of the standings in Formula One isn't just about building the best car, at least not in the way you might think.
Pierre d’Imbleval: Okay, so my name is Pierre d'Imbleval. I'm French, as you can hear it.
Becca DeGregorio: Pierre is the Chief Information Officer at Renault Formula One. He manages all the IT activities within the team, which is a whole operation beyond just the driver and pit crew. I dialed him up at his office in the UK recently to find out how his team is hoping to gain an edge on the competition. Pierre joined Renault when the car maker reformed their Formula One team, back in 2016. Their goal was simple, grow and improve for three years, spend the next three becoming a viable contender for the championship.
Pierre d’Imbleval: We start from the back of the grid in 2016. We were nine in the championship, then six in 2017, then fourth in 2018.
Becca DeGregorio: In 2019, they're in fifth, fighting to defend that position at an intensely competitive level. They're halfway to the top, and the climb from here just gets even harder. The closer you get to the top, the smaller the margin is between the teams, is that what you're saying?
Pierre d’Imbleval: Yeah. Yeah, and the more difficult it is to really catch up the last percentage of performance out of your machine.
Todd Whitney: Where do they go from here?
Becca DeGregorio: Well, it's hard to say because there's so little margin for edging over the competition. The simple answer is build the best car, but the best car for one race isn't the best car for another. Over the course of a season, you'll go from a race like the one in Monza, northeast of Milan, known for long high-speed stretches, to the one in Monaco, where cars are basically racing through the streets. There's a best car for both, but it's not the same car.
Pierre d’Imbleval: If you race at the end of the season with the car that you designed for the beginning of the season, you will probably be stuck at the back of the grid because those cars are just prototype that constantly evolve through the season. All the data that we discuss gathering along those events are actually essential to design a more performant part for the next race.
Becca DeGregorio: That evolution of the car from race to race, it's happening fast too, and without the opportunity to do much physical testing at each track.
Pierre d’Imbleval: In Formula One, one of the main regulation is that we are extremely limited in terms of physical testing. If we want to make sure that a solution or something that our engineers have designed, our manufacturer will actually deliver the expected output on the track.
Becca DeGregorio: This is why what Pierre is working on is so critical to Renault's success. Once you reach a certain bar of excellence in racing, data becomes an essential tool to improve your car, your strategy, and eventually, hopefully, your race time. Renault collects a lot of data during races, everything from tire degradation to aerodynamic resistance. 220 sensors on a Renault car can collect over 50 billion data points in just one race weekend.
Todd Whitney: Then what happens?
Pierre d’Imbleval: To analyze the behavior of the car, we have these two data feeds that goes ... One's locally, track side, and the other one's back to the factories. All of that information becomes available for those engineers. When the car is backed into the garage, there is the possibility to gather even more information because you can plug a cable on the car and gather those data at a much higher frequency such as during the session by the telemetry system. That will refine the analysis of the engineers to prepare for a nicer setup, or based also on the feedback that is given by the driver to the engineers, if he feel comfortable with the car or if he feels too much oversteer or understeer driving the car.
Becca DeGregorio: From that data, a sort of redesign takes place, maybe a change in the steering wheel position or a different tire type from the options Formula One allows.
Todd Whitney: Now, I imagine Pierre's team isn't the only one doing all this.
Becca DeGregorio: Right.
Pierre d’Imbleval: I would rather say that we use fast data compared with big data. To be honest, in our case, the volume of data that we generate seems big. Racing on the track, we are generating approximately 50 billion data points during the weekend, but from a data volume, it's not that massive. What is super important in our sport, it's to get those data as fast as possible. What can make the difference in our sport is how fast you can get the data to start the analysis in order to take better decision, more accurate decision, depending of the speed you get the data.
Becca DeGregorio: Cloud computing is critical. Renault has been adopting services across Microsoft's Azure cloud to help with this. Prior to Azure, what was the team utilizing for data analysis? What did the method look like?
Pierre d’Imbleval: Excel spreadsheet.
Becca DeGregorio: If you didn't catch that, Pierre said, “Excel spreadsheet.” He's only partially kidding. Moving from on-premises servers to the cloud for some of their most computationally taxing work, Renault can be sure the latest systems and technology are powering the millions of simulations they're running, and they can scale up or scale down as their needs change too.
Pierre d’Imbleval: For tire degradation, for instance, or for anomaly detection in our sensors, we use the machine learning algorithm in Azure to help us to reduce the time that our engineers spend to detect what are the sensors in defect by just analyzing not only the measure that the sensor is giving, but also what is the behavior of the sensors next to this one. We teach the model how to decide if that sensor is delivering good measure. That's a great experiment that we had with Azure machine learning that helped us to get better in anomaly detection.
Becca DeGregorio: What's on the horizon for Pierre and Renault? It's continuing to refine their approach, not just to data collection, but how they interpret it. They're continuing to push on better data visualization tools and simulations, and more sophisticated AI that can put the 50 billion data points they're generating each race weekend into context.
Pierre d’Imbleval: It's probably the most technology-driven sport in the world. We are at a moment in the sport that you cannot start the car without having the proper IT solution systems and infrastructure in place to start the car.
Becca DeGregorio: When every team has an IT solution, it's these creative and innovative uses of data that will help Renault edge out the competition and continue their steady rise through the ranks of Formula One. Todd, what do you think of all that?
Todd Whitney: This idea that with the 50 billion data points Renault is working with, it's more about fast data rather than big data, I love it. I can see how it's less about volume than what you do with it, how you interpret it, how you apply those interpretations, and how quickly you do all that.
Becca DeGregorio: Did you find out something similar in the conversation you had?
Todd Whitney: I did. Let's talk about e-sports. People have been competing against each other in video games for decades, but in recent years, competitive gaming has evolved into a full-blown sport, one with its own industry, its own competitive structure, its own star athletes, and even its own tournament broadcast. And fans, lots and lots of fans. Those numbers are only increasing. From 2016 to 2019, e-sports went from having 281 million fans to 436 million fans. Top e-sports athletes sign to organizations that put up teams to compete across various games, from sports titles to shooters to multiplayer online battle arena games, that's MOBA for short. For this episode, I checked in on one of those orgs, Cloud9. They're based in Los Angeles but compete all over the world. While I was reporting this episode, a lot of the organization was in Germany at a world championship. I managed to catch up with someone who's going to be key to Cloud9's success moving forward.
Halee Mason: Hi everyone. I'm Halee Mason. I'm the lead data scientist at Cloud9.
Todd Whitney: What kind of role does data analytics play in competition over at Cloud9?
Halee Mason: With all these different games and professional players competing to try and win various tournaments, my role is interfacing with the team on the performance side and helping them to do their jobs better through utilization of data.
Todd Whitney: As it turns out, a love for gaming is a big part of why Halee is here.
Halee Mason: I've done a lot of different MMORPGs over the years. Those are massive multi online role playing games such as World of Warcraft or Revelation Online, but throughout all of those different games that I've played, I always have played League of Legends for 9 or 10 years now, since early on when the game was released right around the time that beta was launched.
Todd Whitney: I can't stress enough, League of Legends is a big deal. In competitive e-sports and in the wider world of gaming on every level, every day, over 8 million players login and compete. In the context of Cloud9 and the work that Halee is doing, League of Legends is a testing ground for what the future of competition might look like, and the role that data could play in the e-sports landscape moving forward.
Halee Mason: We're using data science and technology to help drive forward the training and performance for our teams. We've started this initially through working with our League of Legends team. This is our initial implementation where we're working to develop early tools and technology that'll help us train more efficiently. A big component of this is just leveraging data that had previously been inaccessible to the team, or just not commonly referenced. When you think about traditional sports, there's a lot of data available for that, but when it translates to e-sports, since these games are electronic, there's even more data available. That's really great, and it's also a challenge because you have to identify, with all of this available data, what's really important. How do you identify the really important insights and start at the beginning with what inputs are available?
Todd Whitney: Halee and other data researchers at Cloud9 collect their team's training and computations and then use artificial intelligence algorithms to make the best informed decisions on teams and tactics. How does she do it?
Halee Mason: We know we have upcoming matches against various opponents and we can really utilize the data to learn about our opponents and their tendencies or ways they operate as a team. Another large component of the types of data analysis that we might do is analyzing the matches after the game. We're using different technology to highlight the areas for improvement. It's not just watching the games from a manual sense, but it's automating insight extraction, because we only have so many hours in a day to review matches. The team often plays up to six games a day during their scrimmage periods where they're training and practicing against other teams. Utilizing that information from the post-game and being able to relay that information to the team so that they can take what they learned quickly and go back into the next match has been really helpful.
Todd Whitney: What are some of the things that would have been hard to recognize or see without the current technological tools that you have to optimize training right now?
Halee Mason: A big one that jumps out to me is trends and patterns in the jungle. Within the game of League of Legends, there's different roles and responsibilities that each player has. One of those is through clearing out jungle camps and aiding in the lanes to gain an advantage during the game, a numbers advantage typically. We've been able to use technology and data from analyzing the paths that junglers take in a more automated way to help us extrapolate those trends that we can then take forward into future games.
Todd Whitney: The ultimate goal is to create a tool that can really work with Cloud9's approach. As with Pierre and Renault, Halee and Cloud9 are working with Azure, that's Microsoft's cloud platform. They're trying to build a constellation of services that'll help them get what they need out of all the reams of data that they're collecting, but that's just what's happening on the backend.
Halee Mason: Then another large way that I interact with the team is through working with the coaches. The coach might say, "This is a new patch. The developers of the game have made a significant change to this champion and I think we need to utilize this champion in our practice this week and make sure we're comfortable playing this champion and we know how it synergizes with our team."
Todd Whitney: Halee and Cloud9 are very much at the beginning of their work. Her role is only going to develop and become more important over time. She told us she's aware of a massive uptake in this kind of work for data scientists, and in time, there'll be working with considerably more sophisticated tools.
Halee Mason: I think part of this one comes through utilizing cutting edge technology. Everybody has this information available, but it's how you use it where you can gain that competitive edge. One of these is through using different machine learning models or even deep neural networks for computer vision to provide that next level of insights for training, where it's not easy to do and it takes a lot of skill and guidance to perform effectively.
Todd Whitney: The end goal is to bring us all together into a tool that the team can use on a daily basis to help them train more effectively.
Becca DeGregorio: I feel like with what Halee is working on for Cloud9 or how Renault's Formula 1 team is working with data, we're just at the beginning of all this. We're already seeing gains, but it feels like the potential is massive for players, coaches, teams, and ultimately for fans.
Todd Whitney: We'll be watching this space.
Todd Whitney: To learn more about all the people and stories featured in this episode, visit Microsoft.com/InCulture. There, you can meet more people from the teams of Renault Formula One and Cloud9, and learn more about the role technology is coming to play in sports all over the world. If you want to see more from the series, follow us on Instagram, @MicrosoftInCulture. In Culture is hosted by Becca DeGregorio and me, Todd Whitney. It's produced by Jordan Rothlein and edited and mixed by Nat Weiner, original music by Angular Wave Research. In Culture is a production of Microsoft in collaboration with Listen, a sensory experience company in New York City.