DraftKings built a machine-learning model in 2023 using customer betting records to identify which gamblers were more likely to respond to promotions by betting – and losing – more, according to a New York Times investigation published Sept. 19.

The reporting draws on internal documents and interviews with dozens of former DraftKings employees, several of whom said they worried the company’s targeting efforts were harming problem gamblers.

How the Scoring Model Worked

At the time DraftKings built the model, the company was spending hundreds of millions of dollars annually on promotional incentives – free bets and bonuses pushed through emails and phone alerts – without a clear picture of which offers actually worked, according to the Times.

The model assigned each customer a score based on betting habits; the higher the score, the more money that gambler was predicted to lose for every promotion sent their way. Jayden Butts, a DraftKings data analyst, was assigned to test the model roughly a year into his job, prioritizing free bets and bonuses for the customers flagged as likely to lose the most, the Times reported.

Butts told the Times the underlying logic amounted to searching for traits indicating a good investment, and that by that financial logic, a problem gambler would qualify. Six former employees said DraftKings has since continued refining its data-science methods to direct promotions toward losing gamblers in ways that encourage more betting, the Times reported.

Separately, four other former employees told the Times that DraftKings stalled or squashed a parallel effort to use similar technology to predict which customers might be developing a gambling problem based on their betting activity. The excerpt reviewed does not include a response from DraftKings.

A Wider Industry Tension

DraftKings makes money when its customers lose money, and the Times frames the model as part of a broader shift in which gambling operators have adopted the same behavioral-data techniques Silicon Valley firms spent years perfecting to keep users clicking. That tension – using predictive data science to maximize promotional returns while an equivalent tool for flagging at-risk bettors reportedly went nowhere – sits at the center of the investigation’s findings.

The reporting lands amid mounting scrutiny of sportsbook targeting practices more broadly, including congressional pressure over FanDuel’s VIP program and a push from the MLB players’ union to ban personalized sportsbook marketing. Whether that pressure extends to DraftKings’ own promotional algorithms, following a quarter in which the company’s revenue already fell, is likely to be the next question regulators ask.

The post Loss Predictions Shaped DraftKings’ Betting Promotions appeared first on ReadWrite.

This post was originally published on this site