Primjena optimizirane diskretizacije zasnovane na minimizaciji unutargrupne varijance radi ocjene inherentnog rizika od pranja novca i financiranja terorizma

In the context of an increasingly demanding regulatory framework and the dynamic nature of the financial market, the accurate and objective risk assessment of inherent money laundering and terrorist financing (ML/TF) risks has become a regulatory imperative. This article proposes a methodological approach based on the minimization of intragroup variance, adapted for the classification of quantitative risk indicators into risk categories through interval scoring. The model allows for the empirical evaluation of inherent risk factors – such as customer risk, product/service risk, geographical risk, and distribution channel risk – based on structured responses provided by supervised entities. A key mathematical challenge within this methodology is the combinatorial complexity arising from the distribution of data across multiple intervals, which requires algorithmic processing and cannot be reliably conducted manually. By applying a permutation-based optimization technique in combination with least squares minimization, the model ensures statistically consistent scoring, significantly reducing subjectivity. Although the method applied in this analysis is not the classical least squares regression, it is based on the same mathematical principle of quadratic optimization. Specifically, the methodology minimizes the sum of squared deviations of observations from their group means, with the objective of achieving minimal intragroup variance. The optimization is performed over discrete group boundaries rather than over a continuous function, making the approach suitable for risk classification and score discretization rather than regression modeling. The method applies a least-squares-type quadratic minimization principle to discretize risk indicators into homogeneous groups by minimizing intragroup variance. Furthermore, supervisory oversight is incorporated into the model, enabling justified score adjustments based on inspection findings. This hybrid methodology is adaptable for implementation in dynamic Excel environments or Python-based systems, making it a practical and scalable solution for supervisory institutions. The model aligns with the evolving requirements of the Anti-Money Laundering Authority (AMLA) and supports the development of data-driven, risk-based supervision frameworks.