Predicting Quitting Points and Designing Interventions for Loss Chasing in Sports Gambling

Loss chasing, the tendency of gamblers to persist or escalate betting in attempts to recover prior losses, is widely recognized as a defining marker of gambling disorder. While prior research has illuminated the psychological mechanisms underpinning loss chasing, most studies have relied on self-reported surveys and interviews that suffer from recall bias and limited ecological validity. In contrast, behavioral tracking data from gambling applications offers an opportunity to objectively examine gambling trajectories in real time and at scale. This project leverages such data to advance understanding of loss chasing behavior across the spectrum of gambling involvement. Specifically, this study aims to solve three interrelated questions:

(1) Can user accounts be classified to identify potential problem gamblers based on historical gambling patterns?

(2) Can deep neural network models predict the quitting point of loss chasing episodes?

(3) What types of personalized interventions can effectively support behavioral shifts from problematic to non-problematic gambling? By integrating machine learning with behavioral science, this research aims to provide actionable insights into gambling harms and inform the design of targeted intervention strategies that are both scalable and individualized.