Quantitative Finance and Risk Management

 

Model Risk Management Lifecycle

Manage and monitor models across users and lifecycle stages

Model Risk Management

MathWorks Model Risk Management solution enables financial institutions to reduce costs while improving model documentation, transparency, and compliance. Using the solution, users throughout the model lifecycle can:

  • Capture repeatable workflows through code generation, automated documentation, and document linking
  • Automate testing and validation for continuous monitoring using customizable key risk indicators (KRI)
  • Scale algorithms, models, and apps both horizontally and vertically
  • Interop with existing infrastructure, tooling, and languages such as Python/R/SAS
  • Focus on issues across the lifecycle with full model lineage and usage reporting

A Language-Agnostic Model Risk Lifecycle

The MathWorks Model Risk Management solution consists of six fully customizable components that support the management of data, models, and documents across the lifecycle. Each component supports integration with existing tools and infrastructure, from desktop to cloud. All lifecycle stages are synchronized through a centralized model inventory that tracks the full model lineage and use.

Model Risk Management Lifecycle

Model Inventory and Repository (MIR)

Manage models and modeling projects

  • Provide centralized access to models and enhance model reusability
  • Manage model workflow and validation projects
  • Inspect models, intermediate results, and audit trail

Stage 1: Model Development Environment (MDE)

Define and develop

  • Explore, develop, back-test, and document models and methodologies
  • Improve transparency and reproducibility of model development
  • Auto-generate model documentation and reporting

Stage 2: Model Review Environment (MRE)

Review and approve

  • Perform independent model reviews on the complete set of model artifacts
  • Interactively perform sensitivity analysis on model parameters
  • Comment and flag any aspects for response and resolution

Stage 3: Model Test and Validation Environment (MTVE)

Perform quality assurance and validation

  • Provide the environment for approved models to undergo preproduction testing and validation
  • Automatically run unit tests and generate test reports for MATLAB and open source models
  • Compare tests of a preproduction model against the currently deployed production model
Model Test and Validation Environment

Stage 4: Model Execution Environment (MEE)

Implement and deploy models

  • Host production models and scale to end users in a secure controlled environment
  • Deploy models onto a production environment without refactoring or recoding models
  • Integrate with existing technology infrastructures and incorporate ModelOps best practices

Stage 5: Model Monitoring Dashboard (MMD)

Monitor, report, and assess

  • Real-time model execution results using a configurable web dashboard
  • Explore data segments and configure alerts and thresholds for automated monitoring
  • Analyze model usage, behavior, and health to determine candidate models for retirement