Data analytics for harm prevention: Lessons learned from a decade of research and practice
Session Title
Session 1-1-C: Using Data for Responsible Gambling
Presentation Type
Paper Presentation
Location
Park MGM, Las Vegas, NV
Start Date
23-5-2023 10:15 AM
End Date
23-5-2023 11:45 AM
Disciplines
Other Social and Behavioral Sciences
Abstract
Abstract: The potential of cheaper storage and compute costs to collect, analyse, and exploit data has been transforming industries for decades – harm prevention in gambling is no exception. Stakeholder expectations are typically high, particularly with the volumes of data generated through tracked or online play, although awareness of the costs involved in analysing data well is typically low. BetBuddy, now part of Playtech Protect, is part of this trend and began developing algorithms to identify players at risk in 2011/12.
This paper summarises key findings and methodologies from a decade of published research and internal analytics including: (1) How to synthesise insights from behavioural markers of risk to develop user-interpretable algorithms to specify risk levels. (2) How to draw on insights from players who self-exclude (and its limitations). (3) Which machine learning algorithm wins when running a transparent horse race to predict player harm. (4) Monitoring changing player population risk in response to the Covid-19 disruptions. (5) Using estimated player risk levels to drive tailored interventions to promote more responsible gambling – as contrasted against untargeted, population-wide interventions.
We close by outlining what we see as the key open questions that data analytics is well placed to address over the next decade.
Implication statements: Many gambling operators, regulators, and researchers use data to inform their decision making. We hope that this presentation will illustrate data analytics’ potential and pitfalls - some avoidable, others less so – and inspire attendees to share their experiences and pursue more work of this kind in the future.
Keywords
responsible gambling, analytics, self-exclusion, machine learning, interpretability, intervention
Data analytics for harm prevention: Lessons learned from a decade of research and practice
Park MGM, Las Vegas, NV
Abstract: The potential of cheaper storage and compute costs to collect, analyse, and exploit data has been transforming industries for decades – harm prevention in gambling is no exception. Stakeholder expectations are typically high, particularly with the volumes of data generated through tracked or online play, although awareness of the costs involved in analysing data well is typically low. BetBuddy, now part of Playtech Protect, is part of this trend and began developing algorithms to identify players at risk in 2011/12.
This paper summarises key findings and methodologies from a decade of published research and internal analytics including: (1) How to synthesise insights from behavioural markers of risk to develop user-interpretable algorithms to specify risk levels. (2) How to draw on insights from players who self-exclude (and its limitations). (3) Which machine learning algorithm wins when running a transparent horse race to predict player harm. (4) Monitoring changing player population risk in response to the Covid-19 disruptions. (5) Using estimated player risk levels to drive tailored interventions to promote more responsible gambling – as contrasted against untargeted, population-wide interventions.
We close by outlining what we see as the key open questions that data analytics is well placed to address over the next decade.
Implication statements: Many gambling operators, regulators, and researchers use data to inform their decision making. We hope that this presentation will illustrate data analytics’ potential and pitfalls - some avoidable, others less so – and inspire attendees to share their experiences and pursue more work of this kind in the future.