AI trading comes into its own during an awful couple of trading weeks for European sports-betting operators.
Partially hidden by the big news about Kindred giving up the ghost in North America, the company’s third quarter results statement also gave a sighting to an important theme from the first few months of the European soccer season.
The company said that its sports margin for the three months to September came in below its long-term average of 9.7% at only 9.4%. Worse was to come after period close, however, with October’s sports margin at just 8.6% which as the company said was “significantly” below the average.
Similarly under-the-radar were the comments from Entain in its third quarter trading update where CFO Rob Wood noted that between half and a third of the 6% decline in pro-forma online NGR was due to adverse sporting results, notably European football, in September. He noted that those “customer-friendly football results” continued into October.
Notably, he went on to say that over the past five years the company had seen only three loss-making weeks on football and two of those three came in October. He said the company estimated the impact from these adverse results in October would be approximately £45m.
“We’re calling it Red October,” says Darren Small, Senior Vice President for Managed Trading Service or MTS at Sportradar. “October was the perfect storm in terms of results colliding with some other trends in football.”
Time added on
Small is able to put some flesh on the bones of what was said by Kindred and Entain. He says that on two successive weekends, and during a midweek round of Champions League games in between, the industry saw about 90% of odds-on favourites winning.
“It wasn’t just that the favourites were winning,” he adds. “We also saw a lot of both teams to score bets coming in as well as the half-time, full-time result. It just pounded the bookmakers.”
Small points out that Sportradar’s MTS – which is now used by over 220 bookmakers both large and small – is in part designed to mitigate against such loss-making weeks.
Specifically, Sportradar has its new product, Alpha Odds, which is designed to use the full AI capability at the company’s disposal to smooth out the margin performance of its operator clients.
It is helped, of course, by AI. “What the system is doing is what would have been done manually 15 years ago with one trader on one match who was looking at the financial data and taking a subjective view as to where they need to put the odds,” says Small.
But now AI has “allowed us to do this at scale and we’re able to run this across all of our soccer offerings,” he adds.
“To be clear, we don’t just let the machine run wild,” Small says. “We have our operational account managers on board to make sure the technology is performing as it’s supposed to. Additionally, we have a trading team that is monitoring every betting ticket as it comes through, again making sure that the systems are doing what they should be.
“And recently we installed a brand-new specialist liability trading team who oversees that technology and makes sure it is configured in the most efficient and appropriate way for each of our clients.”
Efficiency is key when it comes to events in October. “The books that were using Alpha Odds were able to withstand better the sheer weight of favourites winning,” he says.
He says that for the past 12 months, Alpha Odds has consistently driven about 1% improvement on the customers GGR from soccer. “The impact of October was very, very significant,” he says. “We looked at the performance of the Alpha Odds customers inside MTS versus that of the customers who weren’t using Alpha inside MTS.”
He says the technology within the Alpha Odds allowed operators to achieve a better understanding of where their exposure lay, with real-time capabilities offering ongoing risk reduction, something the most skilled traders would be unable to do. “It effectively helps operators to reduce those big exposures while still taking the same amount of business,” he says.
“When it comes to the football results in October, it helped to reduce trading volatility from weekend to weekend because of the understanding it has of an operators’ held exposure and the insight it has on how customers are placing their bets.
“The AI models within Alpha Odds understand that the winning customers tend to do this off the back of their wins and they tend to reinvest, they tend to place a bet in a particular way. So that’s why I think that the Alpha Odds product responded so successfully.”
It leaves the sportsbooks on the MTS platform better prepared to take on the next looming challenge facing the industry – the changing nature of the actual game. As Small points out, a huge factor within the struggles being admitted by even the biggest sportsbook this current football season is the amount of added time for each game.
Time is of the essence
“From the World Cup last winter, when you saw the changes in behaviours of the referees and how they added a lot more injury time, that has translated to the domestic competitions. Injury time has gone up considerably,” says Small.
That matters because, as arguably we have seen in October, it leads to more favourite winning. “The favourites have more time, so they score more,” he says.
“This is true statistically. There’s been more goals scored in the top five leagues in this first part of this season compared to the same period in previous seasons and particularly the season before in a pretty significant way.”
“We’ve seen a little bit of looseness around the way the industry has responded to that,” says Small. “And I don’t think it has responded entirely the way that we think mathematically it will go.”
For now, the models – whether Alpha Odds or those employed across the industry – are unlikely to be adjusted “based on some early-season outliers.” But the odds machines will be keeping a close watch.
