Q&A: PandaScore on esports data, AI capabilities and flipping traditional bettors

Paris is commonly known as the City of Love, but a start-up nestled in the 10th arrondissement of France’s capital is looking to change that moniker to the City of Data.

PandaScore is an esports data supplier, and having been launched in 2015 by CEO Flavien Guillocheau, it is now on a rampage to bring esports to the masses.

A significant strategic shift in recent years to supply esports betting data to operators, coupled with the esports boom in 2020 thanks to the coronavirus pandemic, has seen PandaScore skyrocket into conversations in sportsbook high offices.

Speaking to EGR Technology, Guillocheau details PandaScore’s journey from a one-man team to an AI-powered battleaxe, how lockdown was a silver lining and what the future holds for esports and its much-desired data.

EGR Technology: Firstly, why did you decide to launch PandaScore – what was the desire to bring esports data to the masses?

Flavien Guillocheau (FG): It comes down to my interest in traditional sports and esports. In the NBA, it’s very easy to look at statistics and bet on a match, and when you tried to do the same thing for esports it was not possible, or very difficult.

What was lacking was a good data source, so it started from there. Very quickly, we realised being a data provider was an essential part of the ecosystem and if we wanted to help the esports industry grow, being this missing piece was an interesting spot.

As a data provider you try and help many types of businesses, from media companies, broadcasters, betting companies and so on. A year and a half ago, we pushed a bit more towards the betting product because we had some conversations with bookmakers and we realised they wanted  more than just raw statistics, they wanted odds and probabilities, so that is our top focus right now.

EGR Technology: As a relatively new company, what was the impact of Covid-19 like for PandaScore? Did you feel it worse than most?

FG: It could have been worse for sure. The good thing was that for three months esports took the lead and managed to be quite resilient to the crisis. We were lucky enough to be able to raise money during the crisis. We started raising back in April and closed in July so that was a big achievement for us. I think it all comes down to the fact that esports is resilient. It has shown a lot of growth and opportunity. Since we launched a bookmaker product in December 2019, we had the initial traction during Covid, and the traction continued to grow so it was a good momentum for us.

EGR Technology: Can you talk me through how PandaScore utilises AI to collect data?

FG: You have to imagine when you are one person for the first nine months and your job is to collect a vast amount of data on esports matches, there are hundreds of data points for each game. So, when you have limited capacity, you try and find a solution.

In the beginning I was trying to build a proof of concept around how we can automate data collection and, at the time, AI was not as developed as it is today. I was using a parallel between the way some apps can recognise credit card information via optical character recognition (OCR) technology and applying it to esports matches.

Very quickly, we moved to deep learning models (AI) and we built something more robust. When it comes down to how it works, it is simple. You have a model and that model is trying to answer one question. That question might be: “who is this champion?” and you feed the AI an image and it answers.

We have multiple models that each solve a question. We take the video feeds; we have a first model that answers which data is where. Such as where the items are, where the champions are, what the score is and so on.

From a very high level, we narrow down to zones, and then the zones are analysed by proper models. So, in the end from a video feed to statistics, it roughly lasts half a second and then we can extract 200-300 data points from this.

EGR Technology: When it comes to creating odds – is that done via AI or do you have a team of esports-dedicated traders?

FG: When we have the data, we also have another type of model that answers another type of question around probability, which uses machine learning technology. We will ask the model, for example, what is the percentage of one team winning, and then we have the trading team that will work with those probabilities and make them even better.

The trading team is structured by esports. We have CS:GO experts, League of Legends experts, Dota 2 experts, as well as experts for minor esports. They take the knowledge they already have and look at the odds model to decide whether they are great odds or not. The way I see it is the traders are enhanced by AI.

EGR Technology: You cover the majority of the big esports titles out there but are there any plans to secure data for others – Valorant for example?

FG: Our interest is our customers’ interest. They care about having more esports, because for them, it’s more content. It’s important for us to follow all the new trends, with Valorant being one, so we’re getting ready to launch a Valorant product. There are other esports we have on the menu for next year.

From a technological approach when we launch a new game, we can launch a product quite rapidly with basic stuff and then we can increment from there and in a few months have a very solid product.

EGR Technology: On a more general side, do you see esports betting being truly taken up by the larger operators moving forwards? Entain has already said it will be a core focus, but do you think they will know what they are doing?

FG: We are already offering odds to Ladbrokes in Australia and New Zealand, so we have the proof of what they are saying!

I think one of the big differences is in the quality of the live coverage. Many companies don’t really have the capacity to collect data from all the industry and they will have a hard time in having a good product with a good availability of markets. If you’re thinking about what esports fans want – they want live, they want interactivity and if you’re not offering this it is going to be very difficult to appeal to esports fans.

But, looking at the numbers, it is already big and operators who offer a great esports experience start to recognise it.

EGR Technology: When we talk about the rise of esports betting, is the focus to encourage existing esports fans to transition to betting, or is it a case of convincing traditional bettors to switch to esports products?

FG: In the short and medium term, you have to talk to the esports fans. Esports fans want to bet, and some already bet, but sometimes they bet without money because they don’t know where to bet with money.

Apart from when there is a sports shutdown, it is difficult to imagine sports fans will be on esports. You have to have them relate to esports and understand it first. Sports fans will look at FIFA because they understand it, but it is not the best esport and it’s not as entertaining as CS:GO. Maybe a sports fan will find Rocket League, which is basically football played by cars, entertaining.

Progressively, the sports fan will get exposed and used to esports, but I think it’s hard to expect this move to happen shortly. I think it’s a longer process, but it will come.

EGR Technology: Are there any more partnerships in the pipeline you can mention?

FG: As far as the expansion goes, it is two-way. We want to work with tier-one bookmakers and great names because we think that’s the best way to make the esports ecosystem more known and also to show the whole betting ecosystem what esports really can be.

We also want to work with big platforms that are distributing software to bookmakers because we think that’s also a way to have a reach and to profoundly change the way esports betting is being handled.

EGR Technology: What are your hopes and expectations over the next 12-24 months?

FG: It comes down to our vision and why we are doing things. Since the beginning, the goal was to help the ecosystem grow. I think now we provide that as a company, we’re able to generate a decent amount of revenue and we have good growth. It’s about how we can generate and sustain this growth and have as big an impact as possible.

Our goal is to become the number one esports data company that can help the ecosystem. What I hate is when a customer says ‘maybe I can’t do this, maybe you don’t have the data for this for the kind of experience I want to design’ and that’s something we want to change because our customer should be able to build anything they want, so we should be able to provide the data they want.