Sportradar talks generative AI at investor day with aim of driving profits
The company demonstrated a number of its AI offerings at its investor day on Tuesday (1 April). Sportradar’s CTO, Behshad Behzadi, told the conference the company’s AI technology product base starts with data – something which it has been building out for many years.
“This is a rock-solid foundation. It’s years of optimisation which has gone into this. So it is very hard for anyone else to match it, or very expensive to reach this level,” Behzadi said.
Behzadi said its infrastructure is processing millions of data points in real time and is storing them to further help flesh out Sportradar’s AI models and products.
“We are collecting the events’ data, the betting data that is going through. For a particular bookmaker, based on their risk appetite and their reliability, we optimise the odds continuously in real time,” Behzadi explained.
The goal, Behzadi said, is to maximise operator profits by readjusting their offerings. He stated Sportradar’s in-house AI models would post 11% profit increases.
Sportradar generative AI brain
Behzadi also demonstrated the generative AI model they have dubbed the ‘Sportradar Brain’.
The model, essentially, is connected to the data the company has collected and can execute sports- and betting-related tasks or queries.
The group’s CTO asked the AI model which of three NBA teams had the highest chance of winning at the end of the first quarter in their most recent games. The model responded with individual odds for each team, picking the highest from the bunch.
Sportradar’s generative AI offering can also write sports articles and include charts and graphs.
“We have been showing it to some of our clients, who have been very excited – our media clients particularly,” Behzadi said.
Sportradar computer vision and AI scripts
The company showcased its computer vision software that tracks live games, collecting and storing data.
Sportradar said it is currently collecting data from over half of the live games it covers.
“When a game is covered by computer vision, it’s up to 90% cheaper than if you were covering it with our data collectors,” Behzadi stated.
“It’s not that efficiency that is the most important part – it’s actually the fact that we’re also getting 100 times more data.”
Sportradar has also noted it is using automated AI scripts that write code to further develop its AI offering.
The company has stated AI will improve the productivity and efficiency of its engineers more than 50% by 2028.
Sportradar forecasts 2027 results
The sports data analytics company has forecast it will deliver revenue of €1.7 billion ($1.8 billion/£1.4 billion) in 2027, as well as an adjusted EBITDA of €455 million.
In its full-year results, Sportradar reported revenue of €1.1 billion ($1.2 billion/£927 million), up 26% year-over-year. Profit for the period was €34 million.
Sportradar CEO Carsten Koerl commented on the forecast: “Sportradar is uniquely positioned at the centre of the sports ecosystem. With our leading scale, unparalleled global distribution network and history of innovation we are confident in our ability to continue our strong momentum and deliver tremendous value for our clients, partners and shareholders.”
In M&A news, the company announced a deal to acquire Endeavor’s IMG Arena business on 19 March. The deal saw Endeavor pay upwards of $200 million to offload its sports data business unit.