The algorithm developed by the ANJ reveals an alarming and rising number of problem gamblers

In its 2024–2026 strategic plan, the ANJ has placed the reduction of excessive or pathological gambling at the heart of the sector’s regulation and of its expectations of gambling operators.

It has therefore called on all market actors to fulfil their legal obligation to identify and support problem gamblers by identifying more of them, in line with the size of their player base, and to provide them with better support. This support can take various forms, such as telephone calls, suggestions to introduce moderators or appropriate gambling limits, putting players in touch with charities or healthcare centers, or even closing player accounts, etc.

During the annual review of the ‘Prevention of Excessive or Problem Gambling’ action plans, the ANJ noted an improvement in the results regarding systems for identifying excessive or problem gamblers online. This resulted in a threefold increase in the number of players identified between 2024 and 2025, rising from 31,000 to 89,000 excessive gamblers identified by operators.

However, this figure remains inconsistent with the size of the operators’ player base and prevalence studies, leading the ANJ to ask them to identify even more problem gamblers by strengthening their measures.

In order to make its compliance requests more objective, the ANJ has developed an algorithm to identify problem gamblers within the account-based players (licensed online operators and FDJ and PMU account-based gambling). This tool forms part of the initiative to support operators and has two objectives :

• To help operators offering online gambling services to comply with their customer identification obligations ;

• To enable the regulator to objectively assess the efforts required of operators in terms of identifying problem gamblers and reducing the GGR generated by these players.

However, this algorithm is not designed to measure the exact number of problem gamblers or to estimate the prevalence of problem gambling in the same way as barometric surveys of the general population.

How was this algorithm developed?

The ANJ has access to all online gambling data generated through player accounts held with a licensed operator or with FDJ and PMU. This data is transmitted to the ANJ on an ongoing basis by the operators, enabling it to carry out its missions.

It was primarily from this pool of data that the algorithm was developed from 2024 onwards, alongside scientific literature. In 2025, the ANJ presented this algorithm to operators and organized a consultation.

The algorithm uses 23 risk indicators or criteria to generate a single score for each player. These indicators relate to financial transactions, gambling moderators, gambling activity and frequency, as well as the player’s history. 

Depending on their score, players are divided into four categories: recreational players, moderate-risk players, excessive players and clearly excessive players.

The algorithm’s performance has been validated and measured using the Canadian Problem Gambling Index (CPGI), under the supervision of a scientific committee comprising recognized researchers[1].

While similar initiatives are underway in other countries such as Spain and the Netherlands, the ANJ’s algorithm is the first of its kind available in Europe.

What are the initial results?

The initial estimates are doubly worrying as they clearly show: 

• An upward trend in the number of problem gamblers, which is rising faster than the total number of gamblers. 

In the second half of 2025, the algorithm identified 600,000 players who are highly likely to be problem gamblers, representing 8.7% of the total population of registered players. Of these players, 300,000 are so clearly at risk that their identification by operators is imperative.

Whilst the upward trend in this figure since 2023 is likely due to overall market growth, this does not explain everything, as the number of problem gamblers has risen faster than the total number of gamblers.

• An upward trend in the contribution of problem gamblers to operators’ turnover 

Problem gamblers generated €1.2 billion in GGR, representing 60% of total GGR, a proportion that has been rising steadily since 2023.

What are the applications of this algorithm?

For operators, this is an optional tool that is made available to them in a fully transparent manner and which they can use as a compliance benchmark to assess whether their identification of problem gamblers is at the right level. The ANJ algorithm can be used alongside each operator’s own algorithms.

For the regulator, the algorithm provides a valuable benchmark for monitoring trends and assessing the effectiveness of measures to prevent excessive or pathological gambling – the primary objective of government policy, to which operators must contribute.

As part of the review of the ‘Prevention of Excessive or Pathological Gambling’ action plans in 2027, the ANJ will, for example, be able to compare the number of excessive gamblers reported by operators with the number detected via the algorithm.

In any event, the ANJ expects operators to identify, in the short term, those players who are clearly excessive (around 300,000), and to work towards identifying the entire population of excessive players as identified by the algorithm (around 600,000).

Isabelle FALQUE-PIERROTIN, Chair of the ANJ, said: “The finalisation of this algorithm and its release to operators marks a decisive step for the regulator. It demonstrates the regulator’s ability to develop an innovative and effective tool, designed to closely reflect the actual behaviour of online gamblers. In addition to survey data, the algorithm helps to objectively assess the efforts required to identify problem gamblers, whom operators must take immediate action support. Finally, it appears necessary for this identification process to be carried out at points of sale as well, an objective we have been calling on the two monopolies to pursue since 2024.”

[1] Jean-Michel Costes (Associate Researcher, Research Chair in the Study of Gambling, Concordia University, Montreal, Canada)[1], Amandine Luquiens (Professor, Addiction Department, Nîmes University Hospital, CESP,  Inserm U 1018)[2], Gaëlle Challet-Bouju (Hospital Research Engineer (PhD, HDR), Nantes University Hospital, Inserm UMR1246 SPHERE Team) [3] and Sylvia Kairouz (Professor and Director of the Research Chair in Gambling Studies, Concordia University, Montreal, Canada).