for early access to the latest igaming videos!
and get the latest igaming news first!
A machine-learning model commissioned by the Dutch gambling regulator is designed to spot signs of risky online gambling by studying what players actually do, rather than what they report about themselves.
Researchers at the University of Amsterdam developed the open-source algorithm for the Kansspelautoriteit (Ksa). It examines betting behaviour, gambling frequency and the timing of sessions, while also looking at winning and losing streaks and how players react to them. The system then produces a risk score for each player.
The project covers online gambling broadly and is intended for use by operators as an additional layer in their player-protection systems.
Behaviour Over Self-Reporting
The model is built around activity data. That matters because gambling risk can emerge through changes in behaviour that a player may not recognise or disclose.
The researchers drew on earlier work by Spain’s gambling regulator, the DGOJ, when developing the approach. By making the model open source, the project also gives operators a framework they can incorporate into their own monitoring systems.
But the Ksa is careful about what the algorithm can and cannot do.
A risk score is not a finding that a player has a gambling problem. The regulator does not regard the model as a replacement for an operator’s existing duty-of-care responsibilities. It is meant to help identify behaviour that may warrant closer attention.
That distinction is important. The algorithm can flag patterns, but operators still have to decide what action is appropriate and apply the wider safeguards required under their player-protection obligations.
A Wider European Experiment
The Dutch initiative is part of a broader move by European regulators toward using behavioural data to identify potentially harmful gambling earlier.
France has taken a similar route. Its gambling regulator, the ANJ, has used an algorithm to identify roughly 600,000 account-based players considered likely to be gambling excessively. Those players accounted for about 60% of the segment’s gross gaming revenue.
The two approaches point toward a more targeted form of responsible gambling oversight. Instead of relying only on general warnings, spending limits or player disclosures, regulators are increasingly testing whether patterns buried in gambling data can help operators identify risk sooner.
The Dutch model is not being presented as a finished answer. Its usefulness will depend heavily on how operators interpret the scores and what they do after a player is flagged. The Ksa’s own position is clear on that point: the algorithm is another instrument, not a substitute for human oversight or the broader duty of care.
Source: igamingexpress.com
