Model Validation
ComplyGenics has a dedicated team of professionals specializing in Model Validation. The team uses a combination of data analytics and use test cases to validate model expected results. Additionally, the team has developed a set of proprietary methodologies and tools that facilitate the testing and documentation of the models.
Conceptual Soundness & Documentation Reviews
We evaluate the business objectives for the AML Model to get an understanding of the Model's capabilities and coverage. We then ensure that your AML Model matches the defined business objectives and adjust it if needed. We review and evaluate your documentation starting with the latest Risk Assessment and Model objectives.
Ongoing Monitoring Processes
Ongoing monitoring is crucial to maintain and identify issues and evaluate changes since the last validation. The ongoing monitoring framework should provide continuous feedback about the AML Model that can be used to update the Model as needed.
The following processes should be in place to:
-
Routinely and periodically review the Model performance
-
Identify changes made to the Model
-
Identify assigned responsibilities for ongoing process
We evaluate performance monitoring such as Model effectiveness, Model accuracy, Data accuracy, and emerging risk indicators. Further, we evaluate tuning processes; the monitoring of thresholds documentation and justifying analysis, and trends and KPI are captured and measured.
Outcome Analysis
Outcomes analysis is essential to examine the AML Model’s results and verify that they are accurate and complete. An appropriate transaction period needs to be used to satisfy the AML Model; for example, some models may use a single-day transaction logic while other models may require three or more months of transactions to trigger an alert.
As part of the Data Quality Assessment we perform the following tasks:
-
Assess the quality of data input to the models by conducting targeted testing
-
Assess controls for proper identification and notification of data quality issues
-
Analyze end to end data flow and data mapping from source to the target system, and perform analysis of the ETL (Extract, Transform & Load) processes
-
Assess the quality of data input to the model by testing the consistency of data from source and surveillance databases
