Developing financial crime modeling approaches to more advanced capabilities

Article

Transaction monitoring is key to tracking and analyzing financial transactions to detect and prevent financial crime. Rule-based models are a common approach, but how can more advanced capabilities be used to better the model process? At CeFPro’s Fraud and Financial Crime Europe 2023 Summit, Senior Model Validators from ING discussed how new technologies can advance modeling approaches.Transaction monitoring is the process of tracking and analyzing transactions to detect and prevent financial crime. The approach operates under the premise that unusual or suspicious transactions can highlight crimes such as money laundering and terrorist financing. Therefore, financial institutions must accurately scrutinize transactions to identify potential threats and take appropriate action. It is known that operational costs, time-consuming investigations, reduced efficiency, missed true positives, and reputational risk are all factors as to why transaction monitoring is a challenge.Rule-based models represent a common approach in transaction monitoring, relying on predefined sets of rules, often formulated as “if-then” statements, to make decisions or predictions about the legitimacy of a transaction. While rule-based models have advantages, such as simplicity and explainability, they have several disadvantages. When asking Behrouz Raftari Tangabi prior to the session, “What are the key challenges with regards to rule-based models?” He said:“Rule-based models are not flexible as they rely on predefined conditions, proving challenges in capturing complex patterns. This can be even more challenging in a dynamic environment where adaptability is needed, for example, in fraud cases. Maintaining the effectiveness of rule-based models can be a heavy and time-consuming exercise, especially when the number of rules grows. The modification of rules or adding new ones can also be challenging when there are dependencies or interactions between rules. As rule-based modeling heavily relies

Thank you for visiting CeFPro Connect and reading our latest industry updates. To continue reading more, please create your free account. You'll enjoy the following great benefits:

WHAT'S INCLUDED —