For a Fortune 100 Financial Services firm, Midtown Consulting Group provided project management services in support of new statistical model development of a Value-at-Risk (VaR) model used to calculate margin collateral on derivative trades.
Our client was developing a Parametric Value-at-Risk (VaR) model to support daily re-calculations of initial margin requirements for derivatives in its counterparty credit risk group. In addition to regulatory approval, the model was required to go through internal Independent Model Validation to ensure appropriate statistical methods were employed, inputs and outputs were consistent and of high quality, and that documentation was complete and accurate.
Provide project management assistance to ensure model development efforts met corporate guidelines while ensuring that all development steps are documented in support of model validation efforts. Identify gaps in process and documentation that could impact model approval, and mitigate issues to ensure approval of model.
Midtown Consulting Group provided Program and Project Management services, along with specific regulatory and statistical modeling expertise, including:
- Leveraged past experience developing statistical models and gaining internal Model Validation approval to suggest changes to development
plan and to identify risks.
- Created documentation describing technical features of the project approach, including: description of the business case for the model, outline of the potential model types that were assessed and the selection process for the modeling methodology selected, and created process documentation and descriptions for components of statistical analysis and modeling.
- Defined model validation scope and approach.
- Acted as a liaison between model development teams, regulatory compliance, and model validation teams to ensure key stakeholder participation and ongoing communication.
- Ensured compliance with internal Model Validation guidelines and processes.
- Led efforts to define appropriate system testing and back-testing processes.