New research from Bet Buddy and GTECH: Analysis of demographic and behavioural data from internet gamblers and those who self-exclude

We are very happy to announce that the latest output from Bet Buddy's and GTECH's research collaboration will be published in the Journal of Gambling Studies. This new peer-reviewed paper focuses on self-exclusion as one of the main responsible gaming interventions, and is split into three sections. Firstly, it sets out a three tier model for assessing at-risk gambling behaviours which examines player exhibited, declared and inferred  behaviour. This is an important model that we will be building on in our research projects during 2014 and beyond. Secondly, the paper presents a literature review relating to who self-excludes and whether self-exclusion is effective. Finally, it reports the results of an analysis of the exhibited behaviour of internet self-excluders as sampled from a research cohort of over 240,000 internet gaming accounts self-excluders.

Bet Buddy will be presenting the research outputs during the New Horizons in Responsible Gaming conference in Vancouver in January 2014, hosted by British Columbia Lottery Corporation. Simo Dragicevic, Bet Buddy CEO, will also be participating in a panel discussion titled Player Behaviour Panel Presentation: What Have We Learned from “Big Data” in Responsible Gaming? Does it Change the Way We Play?

Bet Buddy and GTECH have been collaborating on research into responsible gaming since 2011. This latest paper has been co-authored with  Dr. Jonathan Parke, Christian Percy and Aleksandar Kudic from GTECH. Previous research published by Bet Buddy, GTECH and City University London analysed casino data in relation to high risk behavioural markers and was published in International Gambling Studies.

International Centre for Youth Gambling: Behavioral Analytics Article

Bet Buddy has published an article in the latest newsletter of the International Centre for Youth Gambling Problems and High-Risk Behaviours (Fall 2013, Volume 13, Issue 3). The centre is a McGill University Research Centre that is focused on research, prevention, and the training of researchers and professionals concerned with youth gambling and treatment. The article describes the principles and applications of behavioral analytics technology in gaming, discussing how researchers such as the Division on Addiction and gaming organisations, such as Bet Buddy and GTECH, have built an evidence base for by using player data was the study of actual internet player gambling data. The article then explains how this is evidence is now being used by Bet Buddy to enable lotteries and commercial operators to better protect at-risk players through the use of Bet Buddy's technology. We would also like to take the time to congratulate Dr. Jeffrey Derevensky, Founder and Director at the Centre, on winning the 2013 NCRG Scientific Achievement Award, a thoroughly deserved award!

A Brief Re-cap from the NCRG and G2E: Part 2

In the second of our re-caps from the NCRG and G2E conferences we provide a summary from the G2E Panel on Proactive Process: Responsible Gaming Online. The session was chaired by Connie Jones, Director of Responsible Gaming at IGT and included Joachim Haeusler, Head of Responsible Gaming at bwin.party, Hillevi Stuhrenberg, Head of Responsible Gaming at Betsson, and Simo Dragicevic of Bet Buddy.

Simo opened up the discussion with an overview of the evolution of responsible gaming tools over the past decade and outlined the factors that are driving the adoption of new responsible gaming tools. Whilst regulation continues to be the major factor driving innovation and adoption in responsible gaming, there are increasingly more examples of how commercial B2C operators are adopting innovative and new advanced tools to help to protect vulnerable players in the absence of regulation. 

For example, Joachim Hauesler, head of responsible gaming at bwin.party, the world's largest listed commercial B2C internet gambling operator, described how bwin.party are using algorithms to detect problematic gambling behaviour to meet regulatory requirements in the Spanish iGaming market, specifically to support decisions as to whether players should be allowed to increase limits. Also Hillevi from Betsson, the Swedish internet gambling B2C and B2B operator, discussed a new online self-help tool that Betsson is piloting with their players which provides proactive online support from their players who feel they are at risk of problem gambling. The support is provided by an independent treatment provider and is voluntary, and any player information given to the treatment provider is confidential and not made available to Betsson. Both of these are great examples of how the commercial sector is seeing the benefits of implementing more personalized responsible gaming tools to help protect vulnerable players, whilst also building their brand equity and customer sustainability. We think it's very encouraging to see some of the recommendations from our industry expert review paper on CSR in gambling written in 2010 now being implemented in the market today.

Connie Jones then opened up the panel discussion with the audience. Much of the interest and questioning was around how B2C operators were using predictive analytics to better understand player behaviors to make personalized interventions. A topic that was discussed in the previous panel surfaced again too - how should operators best share data on self-exclusion? There appeared unanimous agreement on the need for centralized self-exclusion approaches, with Hillevi highlighting the approach to central self-exclusion adopted by the Danish internet gaming market. Simo also challenged the panel further, asking whether at some point in the future operators should collaborate and share not just self-exclusion data but also other player data, such as deposit and timing limits, to offer a unified responsible gaming platform across all operators for players. Whilst it's far to say this suggestion didn't receive overwhelming support from the panel, we feel it's another 'blue sky' thinking idea, that along with the adoption of universal predictive algorithms that was discussed previously, could one day become a standard practice in future regulated gaming markets.

What does the analysis of internet casino gambling player data tell us about behaviour and risk?

Our research analysing internet casino data has now been published in the latest edition of International Gambling Studies (Vol. 11, No.3), a special edition on Internet Gambling. The paper outlines the phase 1 findings from research undertaken by Bet Buddy in partnership with GTECH and City University London which utilised anonymised player data from a research cohort of 128,788 players from three internet gambling sites licensed in Malta offering internet casino and poker to regulated markets. Whilst the research builds on the methodology adopted by the Harvard Medical School and bwin Interactive Entertainment AG (specifically Braverman and Shaffer (2010)), which analysed live action sports betting internet data relative to gambling risk factors using k-means clustering analysis, our research explores the casino results in new contexts not covered in previous research. To the best of our knowledge this research, along with the Harvard/bwin collaboration (an overview of the collaboration can be found here), is the only peer-reviewed research to be published that analyses actual internet gambling data, and is the only research to analyse casino internet gambling data in the context of risk.

Whilst we discussed some early insights from the research in a previous blog this paper contains the full results and analysis. Our results are analysed in the context of risk factors, game structure, player education and clinical models for problem gambling. For example, we suggest that the analysis of some risk factors, such as loss chasing, could prove problematic when using current gambling screens (such as DSM-IV) in the context of internet gambling. Our results showed that real active money gamblers with the highest intensity and frequency levels gambled predominantly on slots type casino games, in comparison to the most moderate gamblers who preferred table games. We also examined how behavioural analysis and feedback mechanisms can help players to regulate gambling behaviour (for example how medical data is being used to help people make more informed choices - see this TED video). We explore how the opportunities that data analysis offers can help move beyond the traditional applications of data analytics in the gambling industry such as in marketing and risk management, in that applying advanced data analysis in new contexts (e.g. healthcare) can identify individuals who would benefit from proactive intervention or lifestyle changes (McKinsey's Big Data report discusses the application of data analytics in industry in general in greater detail). When the results were analysed in the context of clinical models for pathological and problem gambling (such as the Pathways Model) we found certain limitations. For example, we would have benefited from augmenting our existing internet gambling data sets with new data sets, such as call centre data, which can also be used to help predict gambling behaviours (see Haefeli, J., Lischer, S., & Schwarz, J.(2011)). 

We believe that this research is an important step in furthering our understanding of internet gambling behaviours in the context of risk and player protection and welcome feedback from industry and academic practitioners. If you interested in finding out more about our research and product offerings then please contact us.

Cognitive Bias Modification - Can Playing Games Help in the Treatment of Addictions?

In May 2011 The Economist wrote an article on Cognitive Bias Modification (CBM), a new form of therapy that can effectively treat conditions such as anxiety and addictions without the need for traditional methods such as cognitive behavioural therapy (typically consisting of 12-16 hour talk therapy sessions) or drugs.  How does it work? All it requires is sitting in front of a computer and using a program that subtly alters harmful thought patterns, and is found to be effective after only a few 15 minute sessions.

Researchers are now beginning to explore CBM.  Reinout Wiers from the University of Amsterdam and his collaborators conducted a study to test the application of CBM on 214 patients suffering from alcoholism. Their results showed that a group of patients that were subject to four 15 minute sessions over four consecutive days showed that the patients "approach bias for alcohol had changed to an avoidance bias, on a variety of tests", whereas the control group showed no such changes. In the US, a team from Harvard University are looking to launch a month-long programme that will use smart phones to assess the techniques effects on anxiety.

The idea that human interaction with computers and playing games can help in conditioning behaviours and treating addictions is very interesting, especially when considered in the context of the gaming and gambling industries. Why? Whilst the use of CBM in a clinical context is an emerging field one could argue that gaming designers and operators are masters in the application of CBM. An article in Forbes titled Zynga "Appeals to the Same Psychology as Gambling" sheds some light on this. Jeff Tseng, an analytics expert, states that whilst FarmVille "is not gambling, it’s a similar mechanic,” and that “it appeals to the same psychology as gambling does.” FarmVille's success has been attributed to its game mechanics, in that it has been designed to hook people in to keep returning to the game. It's also very well integrated with social media features which are now part of everyday life. In gambling these principles apply too, although game mechanics are often referred to as the structural charecteristics of a game by researchers e.g. methods for paying and receiving winnings, speed of play, gambling features such as maximum stake allowed, and ambience through as the use of stimulating light. An interesting article from Gamsutra titled 'Ethos Before Analytics' takes a deeper look into game design and behavioural conditioning.

Whilst the continuing uptake of new technologies in the industry, such as behavioural analytics, is helping players make more informed decisions to prevent the onset of problem gaming, CBM could be used alongside traditional forms of treatment to help treat addiction. We think there is an exciting opportunity for the gambling research community and game designers to collaborate and test whether the same game mechanics that are used to make games addicitive can also be applied to developing CBM-type games that can help treat gaming addictions. Although it is somewhat ironic to ask the very game designers who are making games addictive to collaborate in the development of games that can help treat addictions.

Further Analysis of Behavioural Markers for High-Risk Internet Gambling

At the Responsible Gambling Council’s Discovery 2011 conference we presented our research and solutions that help lotteries and operators to develop sustainable relationships with gamblers by helping them to make more informed decisions.  Part of our session was focused on sharing our latest research findings on the analysis of high-risk gambling behaviours.

Braverman and Shaffer (2010), from Harvard’s Division on Addictions, published How do gamblers start gambling: identifying behavioural markers for high-risk internet gambling last year.  The paper is an important research asset as it was the first to analyse actual online gambling behaviour during a gambler’s first month of play to predict gambling-related problems.  Their study of live action sports bettors identified a small sub-group of gamblers (2.8%) from the total research cohort (n = 530) who demonstrated high levels of gambling variability and involvement.  These gamblers were found to be at higher risk than other gamblers of reporting gambling related problems.

The study was of particular interest to us because it was the first to use actual online gambling data to analyse  the first month of play, which allows for the possibility of intervention before gamblers start causing harm to themselves.  We wanted to see whether their results for sports betting would be consistent with other gamblers so we recreated the study using two new datasets, casino (n = 546) and poker (n = 575), using the same methodology as Harvard i.e. analysing the first month of play following registration, using the k-means clustering method and the same behavioural markers.  Our results showed some similarities to Harvard’s, in that we identified in both casino and poker a small-sub group of gamblers who showed markedly different gambling behaviours compared with the others gamblers during their first month of gambling activity, which in our case was highly variable gambling patterns e.g. see cluster 3 in our casino results.

We also observed other similarities with Harvard’s results e.g. the majority of gamblers in the research cohorts demonstrated moderate betting patterns.  There were also some differences e.g. our casino research identified a sub-group of gamblers who demonstrated high levels of gambling Intensity during their first month compared to the other gamblers.  This could be attributed to the nature of the casino games, such as slots and roulette, in that they are more continuous compared with live action sports betting and poker, which could allow for more intensive betting behaviour.  

Further opportunities exist to build on Harvard’s and our research, including extending the number of risk factors and also leveraging alternative statistical methodologies.  Whilst clustering is a useful machine learning technique for dividing data into meaningful groups, it has some limitations.  For example it has trouble clustering data with large outliers, such as skewed non-normally distributed populations such as these datasets.  It is also produces more meaningful and natural clusters with the application of greater numbers of variables and sub-clusters e.g. Experian, the credit rating agency, has used the k-means clustering technique to cluster populations into 45 types and 13 groups using 350 measures (Cameron et al, 2005.  A new methodology for segmenting consumers for financial services.  Journal of Financial Services Marketing, Vol 10, 3, 260-271).  

These findings provide further evidence of different gambling patterns and behaviours relative to high risk behavioural markers amongst gamblers.  We plan to publish the full findings later in the year.  However, when new technologies emerge, such as using behavioural analytics to help gamblers to better self-regulate their gambling behaviours, their usefulness is not always obvious and their uptake is rarely dependent on how well the technology works.  Rather, success and uptake depends on whether the ideas behind the technology spread and diffuse, which is why it is important that the industry has the opportunity to debate these ideas in detail.  At Discovery 2011 there was significant interest and debate across many sessions in how new technologies can help to better protect vulnerable gamblers.  Thanks must go to the Responsible Gambling Council for organising a great platform to enable lotteries, operators, software providers, problem gambling prevention and treatment providers and academics to debate these issues in a very open and collaborative manner.

The Gambling Industry and Academia - Should They Collaborate?

A topic of interest within the gambling industry is the question as to what extent research and services to tackle problem gambling should be funded by the industry.  The topic is not new and has been debated for many years.  It was highlighted in Bet Buddy’s and City University's research paper last year and more recently was brought to attention in Professor Jim Orford’s new book called An Unsafe Bet: The Dangerous Rise of Gambling and the Debate We Should Be Having. 

Professor Orford states that government, service providers and academics are trapped in a consensus view about the benign nature of gambling expansion and are compromised in their ability to seriously challenge gambling expansion.  Orford argues that the independence of the academic community is crucial in areas such as tobacco, alcohol and gambling, however he states that there’s a growing risk that gambling research is being co-opted to serve industry interests.  Whilst some of the Orford's anti gambling expansion recommendations will not sit well with the industry (e.g.”UK based gambling internet sites should be made illegal”) it is a well written and researched book and provides an interesting overview of many issues within the industry.

There is no doubt that some conflicts of interest will always exist in collaborations between the gambling industry and academia.  However does this mean that they shouldn't collaborate?  The world is changing, and fast.  As gambling continues to evolve along a steep technology gradient, new and innovative approaches to research are required to keep up with the pace of industry innovation.  Whilst the rise of internet gambling offers exciting new opportunities for research it also requires more multidisciplinary experience and skills to effectively exploit these opportunities.  For example, B2C and B2B operators are best placed to provide access to players and player data and to advise on the features and technicalities of the vast range of online games they develop.  Specialist software analytics providers have the capability of taking player data and quickly identifying the sub-groups of players whose behaviour differentiates them from the norm.  Academics in gambling and psychology are best placed to validate the research underpinning analytical models and the results from them.  It is difficult to find one organisation that has all of this experience and capability under one roof.

So whilst a collaborative approach appears to make sense how does one overcome the conflicts of interest that exist?  Whilst there is no easy solution, academics and the industry working in isolation will not result in fast progress.  One of Orford’s recommendations is that 10% of industry profits should be directed towards problem gambling prevention and research.  Whilst additional funds will no doubt help researchers, forcing the industry to support research in such a manner may not be conducive to building important industry relationships.

An alternative and approach could be for interested collaborators to develop frameworks with which to build partnerships.  Such frameworks could include a series of principles that each collaborating party signs up to.  For example, one principle could be that academics must be held ultimately accountable for the design of research project aims and for presenting results.  Another could be that industry partners have sufficient consultation in the design of research proposals to enable them to fully apply their knowledge and expertise.  All participating parties could be asked to fund their own efforts in any collaboration project independently.  A more controversial principle could be that accountable academics are not allowed to commercialise the results of research that they were accountable for (although commercialising university research does happen).

As with the introduction of new technology, new research approaches would need to be actively tried to assess how effective they are, and either adapted (we are never 100% right first time) or rejected.  There are now examples in the industry where such collaborations have shed new insight into gambling research that could not have been possible without collaboration.  We at Bet Buddy are doing this too and will be presenting our research, undertaken in collaboration with both industry and academic partners, at the Responsible Gambling Council’s 2011 Discovery Conference.  We need academics to keep innovating and advancing research and we need the industry to support them in doing this therefore we believe that despite challanges to making such collaborations work we will see more of them developing as the industry continues to grow and mature.

Can Loss Chasing be explained by Behavioural Economic Theory?

Research by Xuan and Shaffer (2009) from The Division on Addictions (a Harvard teaching affiliate) sheds interesting light on our perceptions of loss chasing in relation to behavioural economic theory. Xuan and Shaffer analysed the play patterns of 226 online bwin gamblers who closed their accounts due to gambling problems. Their findings state that whilst players experienced increased monetary loss and increased their stake size prior to closing their accounts, they did not chase longer odds.

The authors frame their findings in the context of the work in the 1970s by Kahneman, Tversky and Slovic who analysed behavioural concepts and decision making, with Kahneman and Tversky’s paper in 1979, Prospect Theory: An Analysis of Decision under Risk, arguably one of the most important to be published in this field (Kahneman was awarded the Nobel prize for economics for his work in this field).

The group of gamblers in the Xuan and Shaffer study tried to recoup losses by increasing their stake on events with higher probabilities of winning i.e. they become more risk averse and, therefore, they bet more conservatively. However, the players did continue to bet with a greater stake size, albeit with less risky odds. Therefore whilst choosing less risky odds supports the theory of loss aversion, the fact they continued to bet using a different betting pattern perhaps adds less weight to the theory.  So I have an alternative theory that builds on this.

An additional explanation for the gamblers’ behaviours in this study could be the influence of cognitive biases such as regret theory, self-deception, over-confidence, and the sunk cost fallacy, rather than loss aversion. Jacobsen et al’s (2007) analysis of the influence of cognitive biases demonstrated that they play an important role in the development of problem gambling behaviour.  A study by Shefrin and Statman in 1984 that focused on loss realisation provides interesting insight into why investors tend to hold on to stocks that continue to lose value whilst selling performing stocks. One theory relates to avoiding regret, in that investors may resist realising losses as it is proof that their judgement is wrong. Their research also examined other emotional and psychological factors, such as mental accounting, where professional traders use heuristics, such as never letting their losses reach 10%, as a benchmark or anchor when to sell falling stocks.

Xuan and Shaffer’s paper provides a great insight into the gambling patterns of problem gamblers and their reference to loss aversion to explain gambling behaviours is very interesting and carries weight. In addition to loss aversion, other cognitive biases could have played an important role in the decision making process for gamblers as if they were truly loss averse, one can make the case that they would seek to limit their losses (like Shefrin and Statman’s traders) rather than to continue betting. Whilst many people today think there is very little difference between an investment banking trader and a casino gambler, I appreciate a direct comparison is too simplistic. One thing for sure is that this highlights the complexities involved in trying to get beneath the mind of a problem gambler.