Development and implementation of an algorithm for checking bank customers for compliance with the requirements of 115-FZ: innovative approaches to scoring

in the context of the rapid development of the digital economy and the intensification of cross-border financial transactions, the problem of countering the legalization (laundering) of proceeds from crime and the financing of terrorism (AML/CFT) is of key importance for the stability of the banking system and national security. This article is devoted to the development and implementation of an innovative algorithm for the comprehensive assessment of credit institutions' customers for compliance with the requirements of Federal Law No. 115-FZ. The paper analyzes the limitations of traditional manual and strict-rule compliance systems, as well as substantiates the need for a transition to intelligent, proactive approaches based on machine learning. A detailed methodology for constructing a risk-scoring model is proposed, including the stages of collecting and preparing multidimensional data (customer profiles, transactional activity, external sources), extracting informative features, applying ensemble classification algorithms (gradient boosting) and ensuring interpretability of the results. An idealized experimental approbation was carried out, demonstrating excellent accuracy in detecting suspicious transactions and customers. The significant economic effect of the algorithm implementation is substantiated, which is expressed in minimizing regulatory fines, reducing operational compliance costs, increasing reputational stability and optimizing customer experience.

115-FZ, AML/CFT, compliance, machine learning, risk scoring, financial monitoring, anti-money laundering, gradient boosting, economic effect, digitalization of banks, regulatory technologies (RegTech)