“Sports betting can sometimes be like a galloping horse out of control.” That was the opening line from panel moderator Susan Bala, co-founder of AdvancedCompliance Technology (ACT).
Wonder why there’s so much talk about sports betting at Global Gaming Expo (G2E)? Bala said that $168 billion annually is bet on sports in the U.S. and the global number is close to $5 trillion. That should get anyone’s attention.
That’s also why monitoring wagering behavior, patterns, timing, and the like are so important in ensuring that bettors are confident that outcomes are legitimate.
Kenny White, a former oddsmaker who is now ACTs chief of integrity, said in the mid-2000s, he was asked to help learn why sportsbooks were losing so much money on bets on the Mid-American football conference.
Using old fashioned trial-and-error research, White traced most problems to the University of Toledo and their running back. His work led to the FBI arresting the player. As a result and a surprise to him, he was suddenly considered an expert in fraud investigation.
Today, he certainly is one of the best. Instead of just using newspaper results, he and his company call upon artificial intelligence to identify possible fixing in sports.
In January White and ACT alerted authorities about a UFC match between Alexander Hernandez and Michael Johnson. Hours before the event was to begin at T-Mobile Arena in Las Vegas, the UFC cancelled the fight because of “suspicious betting activity.”
Tim Tenpas is a strategist (a new name for an active sports bettor) with ACT.
“AI is very good on things like looking at timing for reports,” Tenpas said. “What does the timing look like? Where are the bets coming from? When a couple of things are flagged and have suspicious timing, that’s how things come together.”
While AI is very good at monitoring and flagging, a human is still needed to interpret the data, so athletes and others are not wrongfully charged. The panel noted that in another recent UFC fight, the betting also looked suspicious.
However, a deep AI investigation, combined with the team’s expertise, revealed that when one book posted a line (or odds) in error, many other books followed their lead, which led to a flurry of strange wagers. The fighters were totally innocent and AI, with assistance from humans, helped prove it.
“Finding the best ways to make money by exploiting the system is no longer easy,” said Dean Sloves, a founding partner of the Bolt Group. “Moving the money is harder now, with better AML controls (anti-money laundering). The tactics have to change, but even with (AI) agents, there still has to be a human. The sportsbooks are now tied to AI agents. But now it is less profitable to get lines.”
Consider how hard it is to spot problems today. Sports books and online sites offer lines on hundreds of sporting events almost daily. They also monitor the thousands of players, coaches, and officials, not to mention the thousands and thousands of bettors and the sites where they place their bets. Without AI, it would be an impossible task.
But there are still some gray areas. One of the biggest issues is getting all the facts and numbers in one place.
“You have your geolocation data in one place, you have your AML data in another place, and then there is the identity layer,” Tenpas said. “You have your bedding volume data in another place. But today, they don’t really talk to each other. Without that full context, AI can’t make the right picture to explain or to observe something.”
However, they all said AI is improving dramatically daily. Bala said that before AI, “We could maybe detect after the game that there might have been a problem. It was never before the game. Now we can do pregame analysis to determine that there’s something wrong with the fight or whatever the event is before it happens. Now you’re talking about in-game as well, props and all the other types of betting, which is really an expansion of analysis that’s phenomenal.”
All those lessons also apply to the future. In addition to monitoring 700 books worldwide, ACT also does the same for 10 prediction sites. With the phenomenal growth of sports betting in more and more forms, nothing may be more important than the corresponding growth of good AI monitoring.
