Machine learning bolstering integrity fight

Machine learning and AI are playing a larger role when it comes to the evolving landscape of racing integrity, the Asian Racing Conference has heard.

Jack Zuber, Senior Manager – Racing Integrity and Betting Analysis, The Hong Kong Jockey Club, revealed computer models through machine learning are helping to monitor racing’s integrity and investigate suspicious betting trends.

He highlighted starting prices and a horse’s settling position as two important factors when employing machine learning.

“Performance is a good indicator of both betting moves and actual performance, but from an integrity perspective, when the parity between either of these is broken, there may be an integrity risk present – which is exacerbated even further when the parity between the betting move and the performance on the day is upheld,” he said.

“To combat this, we have developed models to accurately predict the current performance of the horse.”

“Starting price is a great indicator of actual performance, but surprisingly, settling position is also a great indicator as well.

“Since 2011 in Hong Kong, if you were to bet to collect the same amount on every single horse that settled within one length of the lead, you would be winning at 10 per cent on turnover, and as such, these are two models that we have heavily developed.”