Algorithms and artificial intelligence rule us. It’s why one party or the other flood your email inbox even though you’ve never given to political candidates. It’s why Facebook keeps serving me ads for STEM-related toys for toddlers. It’s how the betting lines shift in response to on-field events during games.
So where, I’ve long wondered, are the data-driven, machine-learning efforts to ascertain or predict whether a person has or will have a gambling problem?
Finally, an answer: It’s coming. And, also, it’s going to take a while.
Read more from the State of Play column:
- Betting Operators Need To End College Tie-Ins Now Before Lawmakers Do
- Unpopular Opinion: This Year’s March Madness Sucked
- Lifelong Gambler Breaks New Ground In Prestigious White House Appointment
- How The Media Gets The Tarkanian Story Wrong In Its NCAA-Vegas Coverage
- Hollywood Shows Its Vegas Problem Again With ‘Elvis,’ ‘Poker Face’ And ‘Hacks’
- A Gambler’s Love Letter To March Madness
- The Next Mayor Of Las Vegas Refuses To Leave Anything To Chance
- March Madness Maniac Elaine Wynn Offers Some NCAA Bracket Advice
- No Winners When It Comes To Oscars Betting
- It’s Time To Boot Donald Trump From Gambling’s Hall Of Fame
- Annie Duke Has Moved On, And Poker Is Poorer For It
- A Sports Betting Kiosk Inside Your Piggly Wiggly? Bet On It.
- The Mirage To Leave Las Vegas With A Rarity: Some Dignity
- Jerry Green’s First Super Bowl Viewing Like The Rest Of Us
- Would Vegas ‘What Happens Here, Stays Here’ Campaign Happen Today?
- Legal Online Poker, Once All The Rage, Flirts With Irrelevance
- 35 Years Later, We Live In Jimmy The Greek’s Sports Betting And Racist World
- Ohio Sportsbooks Mischief Endangers Legalization Elsewhere
Play+ data plays key role in research development
Still, what doctoral student Kasra Ghaharian has gleaned from a megafile of some 90 million financial transactions for American casino apps and websites between 2015 and 2021, is the sort of groundbreaking work that shows us the way.
In spring 2021, Ghaharian, a senior fellow at the University of Nevada-Las Vegas’ International Gaming Institute, got won the lottery of research raw material. Sightline Payments, the company behind the Play+ cards that have become one of the key ways people fund and withdraw money online, bestowed upon the institute that enormous tranche of data with the hopes that someone there might know how to make sense of it.
“We wanted to understand whether, by utilizing payments data, we could identify potential markers of harm that could lead us to build products that make sure people who exhibit some of these traits receive interventions, restrictions, whatever it may be,” says Jonathan Michaels, Sightline senior vice president strategic development and government affairs, in a joint Zoom interview with me and Ghaharian. “You don’t know what you don’t know.”
Ghaharian, along with the computer scientists and data analysts he roped in from other parts of the university, have done that and much more. The 11-page paper published in the August 2023 issue of the journal Computers in Human Behavior, will undoubtedly be seen one day as foundational.
“A lot of my colleagues struggle to get data, and they might end up doing a little survey or something, so yeah, I was really happy about getting this amount of data,” Ghaharian says. “But there was also a lot of pressure to do this right.”
So far, so good.
New problem gambling data supports past beliefs
Ghaharian set out to see what the number of amounts of deposits over certain periods of time could tell him about a customer’s mindset and behavior. In other words, bursts of spending would indicate a player could be in the throes of a problematic gambling frenzy in which they are losing money quickly and then trying to cover or redeem their losses.
The data used in this research does not identify specific people, just their account numbers and whether they were playing casino games or betting on sports, Michaels says.
Here’s how the data broke down:
- 5% “exhibited the highest deposit activity, averaging almost 15 deposits per week,” showed the largest losses and exhibited the highest number of declined transactions.
- 2% of customers “exhibited a high volume of activity in the number of their deposits and withdrawals, as well as high variability in their deposited amounts.”
- 8% had “transaction frequencies, amounts, and variability” that were higher than the majority of customers but not as high as the two smaller clusters.
- 88% of customers “exhibited patterns of behavior that were not suggestive of any unsustainable or harmful behavior.”
Remember, this is first-of-its-kind research. Ghaharian is trying to create clusters or categories of similar players so he can identify different intensities and styles of problem gambling. He or other researchers will inevitably concoct other categories of cohorts with similar characteristics that become more useful.
The important takeaway here is that some 12% of customers have behavior that could raise a red flag, although only about 3.7% are exhibiting anything that might be considered manic or acute. Ghaharian explains:
“Does that mean 12% are problem gamblers? No. Gambling addiction is a clinical addiction. But we want to build different markers of harm and ask, ‘Where on the continuum do they fall?’ Are there steps that we could or should be taking to make sure that these folks do not go above and beyond their means? Should we be putting, if we see somebody continues to have card declines, a cap of, say, $100 a month? I don’t know.”
Indeed, clear conclusions are hard to come by here except, perhaps, the fact that 88% appeared to be casual players who bet modestly and within their means. That’s a big deal because it empirically provides reassurances that what we believed was true: The share of people with gambling problems is troubling but relatively small.
As with alcohol, sex or any other activity that can become addictive in some cases, most adults are fully capable of self-regulating.
What to do about those who cannot is the most important question the industry faces.
What’s missing in this study?
As a first experiment, the work Ghaharian and his team are doing is enlightening and critical. But there’s so much unknown here that is necessary to begin even considering how to design interventions.
Because of privacy regulations, Sightline can’t fork over everything it knows about everyone whose money it handles. Not only do we not know these people’s names, we also don’t know their ages, genders, geographic locations, or economic circumstances. The dream might be to allow an algorithm that can identify populations likely to have trouble regulating their gambling – young men are often cited as an obvious cohort of concern – and then cross-reference that with, say, their credit score to figure out if they’re playing too much and likely to run into trouble.
There isn’t anything, however, to prevent Sightline’s own data folks from taking the sort of analysis Ghaharian is pioneering to another level. For the moment, though, Michaels wants the independent experts to use the data he can provide to think up ways to respond to people in trouble:
“I could hire Kas for our data-science team, but it was really important to get that academic viewpoint on this on the outset. It adds a lot of credibility to what we’re doing and we want to release it publicly so the world really understands how payments can help identify markers of harm.”
There’s also only so much the industry can do to stop someone determined to destroy their financial lives. The upside of Sightline’s data is that the Play+ method is available on basically every legal app, so in theory that would help build out a net that prevents a troubled player who gets limited or blocked from just going to a different app to play. But that same person could then resort to the illegal market where they are at even greater risk.
Still, it’s an ignition point for tackling a very serious problem using the most obvious tool, the thing that moves gamblers’ money around in this modern world. Says Ghaharian:
“Payments and gambling are inextricably linked.We need to understand payments if we want to understand gambling. It was largely exploratory. No one had really used this kind of data before. We didn’t really have any idea of what potential markers of harm can be derived from payments behavior. So this was definitely a first step in trying to just uncover some of those unknowns.”
Michaels is understandably excited. Since 2021, when the data set ends, there have been several states that have added online and mobile gambling options. It took six years to amass 90 million transactions via Play+, but there’s got to be exponentially more data now, right?
“Don’t give Kas the shakes like that,” Michaels jokes. “As this continues to accelerate, there’s a lot of opportunity to ensure people play responsibly and to help people who do not.”