Q&A - Sportradar's Werner Becher on Simulated Reality

15 Apr 2020

By: Sport Industry Group

Suspensions and cancellations have been dotted across the sporting landscape as a result of the coronavirus pandemic, but in order to fill the void sports data intelligence company Sportradar launched Simulated Reality - an AI-driven product drawing on historical data and statistical output to simulate matches reflecting team form and normal match play.

Sport Industry Group spoke to Werner Becher, Managing Director - US Betting, Sportradar about the new initiative:

How have the first couple of weeks of Simulated Reality gone?

We have been delighted with how Simulated Reality has performed over the first two weeks since launching. It has exceeded our expectations and we have received some very positive feedback from our partner operators and their customers. There have been some interesting results in the top tier games in England, Spain and Germany and football fans have really engaged with the product. More operators are committing to offering it, so we expect to see even more interest over the coming weeks.


What sort of data is being used to create the simulations?

Sportradar has gathered millions of data points over the past 20 years and as Simulated Reality is completely AI-driven and algorithm-based, we have inputted all these into the AI systems we use. This has meant we have been able to create a product that completely aligns with what fans would experience when watching and betting on a real game.

It’s the same teams playing and on the same day and time the games were scheduled, with all the betting options normally offered on a matchday. 


How much data does it take to simulate a full match?

In a normal sporting scenario, we collect data from thousands of real games before our AI then analyses the data coming in and calculates the best markets and odds. With Simulated Reality we generate the data ourselves from our AI engine which is something completely new and unique to the industry.

All our historic and statistical data from more than a decade’s worth of sporting events helps to model an immeasurable number of data points. We provide the AI engine with all the inputs, categories, and filters and within the Simulated Reality product it then selects the best scenes which fit any given moment within a live game. We need to train the AI engine to do this as it can’t be done manually. This quickly creates a vast number of unique game situations, outcomes and gameplays to generate similar but completely unpredictable new experiences for football fans.


How does Simulated Reality take into account in potentially unquantifiable factors like team form?

The AI engine factors in a huge number of different considerations, including team form and fixture congestion. It searches for historic games with similar conditions and odds and, depending on the individual game, then lets the random number generator (RNG) decide which sequences to choose and show next. The engine recognises appropriate happenings and orders, but they are completely unpredictable from a consumer perspective. Simulated Reality gives us the opportunity to model how real games would take place as they would happen in real stadiums, which we think is incredibly interesting for fans.


Was this product conceived in response to the current sporting cancellations? Or is this a product that has been on the cards for some time? 

We had been considering concepts like this for a while, but the time was right to launch it now given the lack of live sporting events and the challenges the sports betting industry is facing. We recognised we had the data, technology and expertise to scale this quickly and create a unique product that provides football fans with an entertaining product and helps our partners generate additional revenue.


You seemed to turn this product around impressively quickly - can you give us any insights about how this was done?

It has been an incredible effort by our different teams to get this product ideated, developed and delivered to market in just a few days. The idea was first conceptualised and fully mapped out on Wednesday 25th March by some of our Austrian product team. The next day we held management discussions where we further developed the concept and held meetings with our AI teams. On the Friday we agreed we would have a prototype ready by 1st April to test, tweak and optimize ahead of launch on Saturday 4th April.

We have 2,500 employees working across multiple global locations, so with everyone currently working remotely it made it even more challenging to create and deliver Simulated Reality in such a short space of time. We had to standardise the software processes, develop the product, test it across every parameter, integrate it into own data feeds and launch it in five days.


Will Simulated Reality continue even when live sport returns?

We see Simulated Reality as a new category within sports betting. We wanted to provide the market with an innovative, relevant product that is exciting, unpredictable and fair for bettors. We believe they will really enjoy the product and while we expect it to be particularly popular at the moment with very little sport on, we see big potential for the future as well because the demand has been so strong.


What other sports can we look forward to seeing get the Simulated Reality treatment in the coming weeks and months?

We started with the three major leagues in England, Germany and Spain, and we will be significantly extending our football coverage in the coming weeks. We’ve also launched cricket and offer Simulated Reality T20 matches.

Over time we will add tennis, basketball, handball and many others to the product range. The AI engine can instantly provide us with more games, so this is a truly scalable product and one that is a major priority for Sportradar over the coming weeks and months