Esports Insights: PandaScore Interview

This week in our Esports Insights series we speak to the CEO of AI esports data provider PandaScore, Flavien Guillocheau (@fguilloc). We chat with him about how AI plays a role in esports and what getting him excited for the future of esports data and technologies.

Could you tell us a little about yourself?

I’m Flavien, CEO at PandaScore; I like video games, technology and dogs.

Do you find time to play many games yourself?

Usually no, but there is a few times a year where I start playing some video games, it’s kind of my holidays. Right now i’m emptying my inbox during WoW Classic queue time, not enough to get high level though… But I played a lot in the past, especially Civilization, LoL, and Total War.

How long has PandaScore been running and what’s changed over time?

I created PandaScore 4 years ago with my co-founder, Jonathan. We’ve had the classic startup journey with ups-and-downs, growing from 2 co-founders to a 40-person team. We raised money a few times to fuel our growth but the vision has stayed the same. We want to help the esports ecosystem grow; our approach to this is to provide data to businesses so they can build their product without having to worry about sourcing quality data.

Recently we’ve been focusing more on bookmakers as they have a strong need in data and odds, but are underserved. We are also working with professional teams, media companies, fantasy apps – anyone that needs data on professional esports events.

You have a wide range of clients using PandaScore’s data. How does this vary from organiser to team? Is anyone doing anything crazy?

This is the great part about being a data provider, there is so many different things our users are building on top of our API. Each customer is unique, some media companies are building statistics website where you can access quickly to historical information or live stats.

Teams are building tools for scouting and for their social media content, there are fantasy betting apps which are using us to build their scoring system. And of course bookmakers, which are using data to give more contexts to punters and our odds for their pricing.

Also it’s very exciting to get the first look on new product coming up as we work with plenty of startups. There are very exciting partnerships that we’ll be announcing soon.

With your recent launch of PandaScore Odds, have you seen the uptake so far?

Yes, definitely. In the past we always had a strong inbound interest from bookmakers because the data they had access to was incomplete, lacked in-play markets, and was inconsistent from one event to the next. That’s why we decided to build an odds offering on the same technology we’ve been using to get live stats for the last 4 years.

Also the ecosystem is more mature these days. A couple of years ago bookmakers just wanted an esports product, so they reached out to non-specialists to check the box. Now they realize that esports is massive. They’re looking for esports specialists that have built a product dedicated to the esports crowd. And that’s us ?

What advantages do you see AI having over traditional for odds generation?

AI is revolutionizing so many parts of betting and esports. Anytime there is dull and repetitive task such as collecting data points, AI will be the best solution. That’s why we use AI for collecting massive amounts of data from Video Recognition. The precise technical description would be Computer Vision with Deep learning neural networks and not OCR as some would state.

Some critics have doubt on the accuracy of AI, that’s most of the time not true. If you look at autonomous cars like TESLA autopilot, there are way less errors per kilometer than humans. If we can trust AI to drive a car for us, we can certainly trust it to collect live esports stats! But since this technology is obscure for many people I understand the skepticism. In the end our Computer Vision AI allows us to collect more than 400 data points in 700ms. You can try yourself at home to compete ?‍?…

The other big advantage of AI is its capacity to handle massive amounts of data and to do inferences from it. A few years ago we were talking about “Big Data” well when you apply Machine Learning networks to our vast amount of data collected with our AI technology you can only get the best predictions. For a game like League of Legends, we have the health point, items or the coordinates of every players at any given second but also for the past 5 years of pro games. Imagine how easy it is to predict where and when the next kill is going to happen.

On top of that, computation take few hundred milliseconds so it’s blazing fast.

Apart from AI, what emerging technology in esports are you most excited for?

As a technical person I look a lot at new things which can change esports and more broadly our world. AI, blockchain technology, ray tracing… but to be honest I think I’m still excited about the development of the older technology in our industry like live streaming. This has created an amazing proximity between players & fans. I think there are still so many amazing innovations coming to live streaming, like in-stream betting for instance.

What are the next esports titles on your roadmap?

We were in a phase where we wanted to consolidate a deep offering so we decided to develop the big four first: League of Legends, Dota 2, CS:GO and Overwatch. Now we are expanding our coverage aggressively and will continue to do so.

Our next games have not been announced publicly yet but if you’re interested in betting on PUBG or Rocket League you should talk to us. After that we’ll look at other Battle Royale games, fighting games, and FPS. The next 12 months will be exciting.

The PandaScore office is giving a blank cheque for a team lunch. Where are you going?

We are a french company, lunch is something we value a lot already, but I guess a 3-star Michelin would not be a bad idea… ?‍? In reality, something more like Five Guys would be my choice. Two milkshakes and a bacon cheeseburger if you’re wondering.