Artificial intelligence isn’t yet ready to push its fleshy overlords out of the trading room just yet. But AI sports betting improvements are coming on the bookmaking side at a rapid pace.
One of the main areas of improvement is understanding the quirks of assessing anything involving humans.
AI is tasked with identifying patterns quickly but currently struggles in the peripheries where the human experience is often defined.
It knows you bought those Taylor Swift tickets. It doesn’t know you did it to impress someone who dumped you.
And it doesn’t yet get sports. Or, more fairly, fallible humans.
The code knows ‘NUMBER 5 has averaged 17 points against DENVER HOME TEAM over the last five years’ but doesn’t realize he had a bad Cobb salad after shootaround, his life partner just bolted, or his Apple+ pitch just got rejected.
AI knows ‘Quarterback, Team X has a 47% red zone completion percentage against ARIZONA HOME TEAM’ but can’t account for the temper tantrum he just threw when his wideout just dropped a potential touchdown pass.
Human oddsmakers can do this. They do this now. Computers can’t. But the computer is learning to take these vagaries into account.
Andre Zammit, Tipico’s US Sportsbook vice president told Gaming Today.
I’m sure they will. Yes. I’m sure they will pick up these abnormalities. I’ve spoken with individuals who think that. Baseball, they’re constantly monitoring the weather as a good factor. So if that changes, or let’s say for example, being over here with the high altitude in Colorado, how does that affect the pitch and the outcomes of the pitch?
“It’s already … towards there, but it’s not there yet.”
NASCAR signed a deal with nVenue in August of 2023 to concoct AI-generated in-race micro-bets.
AI Needs to Learn How Quirks Affect Betting
It’s not just the human element, Zammit noted. It could be atmospheric. Either way, unusual betting patterns on innocuous markets are often a sure sign that a human needs to take a look at what’s happening on the field.
“You always need to ping into certain abnormalities and see, ‘OK, why would this have happened?,'” Zammit posited. “At this stage, I don’t think the data is there yet to really gauge that. So even for us, we have all these automated models for pricing, but you still need that human element to really see why is there traction on this particular selection.
“Let’s say again, going back to the baseball, why is all of a sudden someone betting on every single pitch, someone to maybe hit a home run on the next one? Has the weather changed? Has the pitcher had some injury or anything along those lines? The data is not there yet to automatically pick those, and that’s why the traders would step in and do their private investigation analysis.”
Zammit said the amalgam of granular information available both for bookmakers to construct bets and sharps or dabblers to exploit them increases constantly.
“We focus on AI in-house as well. We try and build AI models, algorithms, the industry’s already quite rich on it,” he said. “So taking basketball as an example, I had products pitched to me, which is like every single player for the last 10 years, where they hit a 3, where they missed a three, so on and so on.
“So all this data helps us big time in creating these micro markets because the data’s so rich and the models continue learning game after game. It’s incredible the state of it, but there’s still a long way to go as well of how to convert that data into betting markets.”
And who or what oversees the outcome.
About the Author
