NBA basketball players, most notably Terry Rozier, were arrested for illegal sports betting. Rozier allegedly told associates that he would underperform during a game that he sat out of early. The associates made outlier bets that triggered cheating detection systems. Dian Zhang and Ignacio Calderon, for USA Today, reported on the usage of statistical models:
“When you do the odds compiling, you have a predicted model for how you expect the game to go,” said Chris Rasmussen, who teaches sports integrity at the University of New Haven and has spent years investigating sports betting fraud for the World Lotteries Association.
Based on the data behind the teams and players in the game, the model expects certain points for those players and predicts “expected behaviors.” When real-world betting behavior starts to deviate from the model’s prediction, that’s when “we are starting to look,” Rasmussen explained. “Why does it deviate, and how much does it deviate, and what’s going on?”
The key is to figure out regular betting patterns and deviate within reason. Avoid the outlier stuff, because of course a concentrated set of bets for a few hundred thousand are going to trigger the systems.
These people need a data guy. Or maybe the good cheaters already have a data guy, and that’s why no one hears about them. See? Math is useful.