Applying Data Science to Behavioral Analysis of Online Gambling
Purpose of Review
Gambling operators’ capacity to track gamblers in the online environment may enable identification of those users experiencing gambling harm. This review provides an update on research testing behavioral variables against indicators of disordered gambling. We consider the utility of machine learning algorithms in risk prediction, and challenges to be overcome.
Disordered online gambling is associated with a range of behavioral variables, as well as other predictors including demographic and payment-related information. Machine learning is ideally suited to the task of combining these predictors in risk identification, although current research has yielded mixed success. Recent work enhancing the temporal resolution of behavioral analysis to characterize bet-by-bet changes may identify novel predictors of loss chasing.
Data science has considerable potential to identify high-risk online gambling, informed by principles of behavioral analysis. Identification may enable targeting of interventions to users who are most at risk.
KeywordsMachine learning Player tracking Loss chasing Self-exclusion Problem gambling
This work was supported by the Centre for Gambling Research at UBC core funding from the Province of British Columbia and the British Columbia Lottery Corporation (BCLC), as well as a research grant from BC Ministry of Finance to LC. XD is supported by a 4-year fellowship funding from UBC. LC receives funding from the Natural Sciences and Engineering Research Council (Canada).
Compliance with Ethical Standards
Conflict of Interest
Luke Clark is the Director of the Centre for Gambling Research at UBC, which is supported by funding from the Province of British Columbia and the British Columbia Lottery Corporation (BCLC), a Canadian Crown Corporation. The BCLC and BC Government has no constraints on publishing. LC has received speaking or reviewing honoraria from Svenska Spel (Sweden), National Association for Gambling Studies (Australia), National Center for Responsible Gaming (US), and Gambling Research Exchange Ontario (Canada). He has not received any further direct or indirect payments from the gambling industry or groups substantially funded by gambling. He has received royalties from Cambridge Cognition Ltd. relating to neurocognitive testing. XD and TL report no conflicts of interest.
Human and Animal Rights and Informed Consent
This article does not contain any studies with human or animal subjects performed by any of the authors.
Papers of particular interest, published recently, have been highlighted as: • Of importance •• Of major importance
- 5.Griffiths MD. Internet gambling, player protection, and social responsibility. In: Williams RJ, Wood RT, Parke J, editors. The Routledge international handbook of internet gambling. London: Routledge; 2012. p. 227–49.Google Scholar
- 7.• Dwyer DB, Falkai P, Koutsouleris N. Machine Learning Approaches for Clinical Psychology and Psychiatry. Annu Rev Clin Psychol. 2018;14(1):91–118 Excellent primer on the history and application of machine learning in psychiatry and clinical psychology, including methodological decisions such as feature selection and cross-validation.CrossRefPubMedGoogle Scholar
- 14.•• Percy C, França M, Dragičević S, d’Avila Garcez A. Predicting online gambling self-exclusion: an analysis of the performance of supervised machine learning models. Int Gambl Stud. 2016;16(2):193–210. Machine learning analysis of self-exclusion in the European GTECH dataset. Compared multiple techniques including logistic regression. A random forest model achieved highest performance (AUROC = 79%) in predicting problematic gambling.CrossRefGoogle Scholar
- 15.•• Haeusler J. Follow the money: using payment behaviour as predictor for future self-exclusion. Int Gambl Stud. 2016;16(2):246–62 The first study to consider online financial behaviours (e.g., amount and number of deposits and withdrawals) in predicting self exclusion. Used artificial neural networks as a form of machine learning to show that payment behaviours achieve a classfication rate of 72%.CrossRefGoogle Scholar
- 16.•• Luquiens A, Vendryes D, Aubin HJ, Benyamina A, Gaiffas S, Bacry E. Description and assessment of trustability of motives for self-exclusion reported by online poker gamblers in a cohort using account-based gambling data. BMJ Open. 2018;8(12):1–8 Reports behavioral tracking over a 6-year period of 1,996 online poker players who self-excluded from the winamax platform. Four machine learning models displayed modest performance differentiating gamblers by their stated reason for self-exclusion (gambling problems versus commercial reasons).CrossRefPubMedPubMedCentralGoogle Scholar
- 21.Tom MA, LaPlante DA, Shaffer HJ. Does Pareto rule Internet gambling? Problems among the “vital few” & “trivial many”. J Gambl Bus Econ. 2014;8(1):73–100.Google Scholar
- 22.• Ivanova E, Magnusson K, Carlbring P. Deposit limit prompt in online gambling for reducing gambling intensity: a randomized controlled trial. Front Psychol. 2019;10:1–11. Randomized controlled trial in online slots gamblers, comparing gambling losses in groups who received a limit-setting prompt upon registration, before or after their first deposit, or no prompt (> 1000 per group). Prompted groups were more likely to set limits but did not differ in subsequent losses over 90-day follow-up.CrossRefGoogle Scholar
- 25.• Cerasa A, Lofaro D, Cavedini P, Martino I, Bruni A, Sarica A, et al. Personality biomarkers of pathological gambling: a machine learning study. J Neurosci Methods. 2018;294:7–14 One of the first studies to apply machine learning to classifying pathological gamblers versus healthy controls, showing 77% overall accuracy using the Big Five personality variables.CrossRefPubMedGoogle Scholar
- 31.•• Leino T, Torsheim T, Pallesen S, Blaszczynski A, Sagoe D, Molde H. An empirical real-world study of losses disguised as wins in electronic gaming machines. Int Gambl Stud. 2016;16(3):470–80 A Norwegian study looking at trial-by-trial behaviour following “Losses Disguised as Wins” in land-based electronic gaming machines (EGMs). LDWs increased the likelihood of continuing betting compared with full losses.CrossRefGoogle Scholar
- 34.Auer M, Malischnig D, Griffiths M. Is “pop-up” messaging in online slot machine gambling effective as a responsible gambling strategy? J Gambl Issues. 2014;(29):1–10.Google Scholar
- 35.Auer MM, Griffiths MD. Personalized behavioral feedback for online gamblers: a real world empirical study. Front Psychol. 2016;7(NOV):1–13.Google Scholar
- 36.• Wohl MJA, Davis CG, Hollingshead SJ. How much have you won or lost? Personalized behavioral feedback about gambling expenditures regulates play. Comput Human Behav. 2017;70:437–45. Intervention study in casino gamblers playing on a loyalty card, who received personalized expenditure feedback. Those gamblers who underestimated their losses showed reduced gambling over 3 month monitoring.CrossRefGoogle Scholar
- 38.Wohl MJA. Loyalty programmes in the gambling industry: potentials for harm and possibilities for harm-minimization. Int Gambl Stud. 2018;18(3):495–511.Google Scholar
- 40.Auer M, Reiestad SH, Griffiths MD. Global limit setting as a responsible gambling tool: what do players think? Int J Ment Health Addict. 2018:1–13. https://doi.org/10.1007/s11469-018-9892-x
- 42.PricewaterhouseCoopers. Remote Gambling Research Interim report on Phase II [Internet]. 2017. Available from: https://about.gambleaware.org/media/1549/gamble-aware_remote-gambling-research_phase-2_pwc-report_august-2017-final.pdf. Accessed 8 July 2019.