We helped a payment service provider create a more efficient transaction monitoring system.
The Challenge: Reduce false positives in transaction monitoring models.
The Approach: Assess the transaction monitoring framework and implement improvements in the Systemic Integrity Risk Assessment (SIRA).
The Solution: Redevelop the anti-money laundering and financial crime aspects of the SIRA.
Client Impact: Reduce false positives and increase operational efficiency.
Learnings: Address more than just the symptoms by applying a holistic approach where the SIRA, customer segmentation, detection methods and monitoring form a consistent framework.
Transaction monitoring systems are essential for financial institutions (FI) and payment service providers (PSP). They help to detect fraudulent customers who engage in criminal activities or are laundering money through the FI’s or PSP’s payment systems.
Unfortunately, transaction monitoring models often produce many false positives; namely, legitimate transactions that are defined as suspicious or risky through the monitoring process. As a result, the operations teams at FI’s and PSP’s often waste valuable time and resources analyzing and properly documenting every alert, including the many false positives.
Our team at Amsterdam Data Collective (ADC) followed a two-step approach. Firstly, we assessed the soundness of the entire transaction monitoring framework. This assessment included the risk identification process as part of the Systemic Integrity Risk Assessment (SIRA) through business rules and detection models. It additionally focused on the alert handling process by the operations team.
Secondly, we implemented improvements that were found in the Systemic Integrity Risk Assessment. Throughout this process, a focus was placed on improvements that were visible, had immediate benefits, and could be delivered quickly.
Additionally, there were three roles necessary to execute this project.
An expert in transaction monitoring, anti-money laundering, and financial crime, as well as the associated laws, regulations and guidelines.
A risk manager with experience in Systemic Integrity Risk Assessment.
A data scientist who understands the ins and outs of business rules and machine learning models in the fields of anti-money laundering and financial crime.
Phase One (six to eight weeks):
To begin, the team at ADC wrote a report on the current transaction monitoring framework. This included both findings on the workings of the framework and recommendations to effectively design it as an end-to-end risk management framework. This included risk identification and scenario setting via the SIRA, risk control actions focused on money laundering detection rules and models, and finally the impact on risk monitoring.
Phase Two (eight to ten weeks):
The next step was to work with the customer operations and compliance teams to redevelop the anti-money laundering and financial crime aspects of the SIRA. This included mapping the existing landscape of business rules and detection models to the individually identified risks.
In addition, during Phase Two, ADC and the compliance team developed a new methodology for customer segments and transaction profiles. This allows business rules and detection models to better target risky customers.
Impact for the Client
Following the project, we expect to achieve a 25% improvement in both the reduction of false positives and increased operational efficiency within the alert handling team. As a result, our client will save both time and resources by reducing the overall number of false positives their operations team analyzes and documents.
Learnings for ADC
This project demonstrates how looking solely at inefficient models or business rules will at best address the symptoms. However, this will not improve the efficiency and effectiveness of the entire transaction monitoring framework. As a result, it is better to take a holistic approach where the SIRA, customer segmentation, detection methods and monitoring form a consistent framework to provide long-term improvements.