Researchers at the University of Amsterdam (UvA) have developed a model to estimate risky behaviour among online gamblers. The gambling risk algorithm uses actual playing data rather than relying only on player surveys. The research was commissioned by the Dutch Gambling Authority (Ksa) and funded through its Addiction Prevention Fund (Verslavingspreventiefonds or VPF).

• The machine-learning model looks at how players actually gamble online. It analyses betting amounts and frequency, including patterns such as playing at night on several consecutive days. The model covers different forms of online gambling and calculates a risk score from these patterns.
• Player reactions to winning and losing streaks are also included in the analysis. This allows the gambling risk algorithm to identify behavioural patterns that could point to increased gambling risk. The researchers drew on work by Spanish gambling regulator DGOJ when developing the model.
• UvA PhD researcher Charles de Leau developed the algorithm together with professors Reinout Wiers from Psychology and Johan Bollen from Computer Science. The research was carried out for the Ksa. The code and methodology have been made publicly available as open source.
• Gambling operators can use the gambling risk algorithm as an additional tool when monitoring player behaviour. The Ksa says its use does not guarantee that operators meet their duty-of-care requirements. It should therefore be used alongside other player-protection measures.
• The Ksa describes the model as an independent and transparent way to support earlier detection of risky gambling behaviour. Operators remain responsible for deciding when player behaviour requires intervention.

Please find more news here.