International Gambling Studies, Vol 16 Issue 2, April 2016

The paper analyses the performance of supervised machine learning models in predicting online gambling self-exclusion.

Categories

This study analyzed behavioral data from over 25,000 online gamblers across six European countries to identify predictors of voluntary self-exclusion—a key responsible gambling (RG) tool.

Key Findings:

• Behavioral Indicators Matter More Than Monetary Ones: Frequent use of RG tools (e.g., limit-setting, prior self-exclusions), multiple payment methods, and playing a variety of game types were stronger predictors of future self-exclusion than monetary intensity (e.g., amount bet or lost).
• Cross-Country Generalizability: Machine learning models trained on behavioral data from one country successfully predicted self-exclusion in others, suggesting these indicators are broadly applicable.
• Younger Gamblers More Likely to Self-Exclude: Age was inversely related to self-exclusion likelihood, consistent with prior research.
• Slots and Impulsivity: High deposit frequency per session and preference for slots were associated with higher self-exclusion rates, potentially reflecting impulsive behavior.
• Ethical and Practical Implications: The findings support the use of behavioral analytics for early intervention, while emphasizing the need for ethical safeguards.

These insights support the development of cross-jurisdictional, behavior-based RG tools that can identify at-risk players without relying on financial thresholds. For full details, please refer to the complete document.