Ongoing Model Performance Monitoring of pricing models across various asset classes for a large US Bank

Client

Large US Bank

To provide a top-tier US bank with ongoing performance monitoring of 50+ pricing models across Rates, Equity, Forex, Commodity, Credit and CVA asset classes.

• Using process standardization and automation techniques, developed an efficient and cost-effective solution to monitor models on a quarterly basis
  
• Market data used is no older than the quarter prior to the assessment
  
• Tests during monitoring include:
  
  • Convergence and Stability Tests (NPV and Greeks)
    
  • Consistency Tests (parity relationships, model limits, accuracy tests)
    
  • Stress Tests (large volatilities, negative rates)
    
  • PnL Attribution (30 business reports and regression tests)
    
• Most tests have well-defined numerical outputs that are compared against predefined thresholds (Green/Yellow/Red)
  
• Monitoring process includes documentation/reporting of the results
  • FX: Cross-Currency Swap Model
    • FX Spot Rate Sensitivity
      
    • DV01
      
    • Consistency with Single-Currency Swap
      
  • Credit & SP: CDS Pricer
    
    • Monotonicity of the Price on the Hazard Curve
    • 30 days PnL (R2)
      
    • Spread01
      

• For equity and FX models, developed common python frameworks that allow running/monitoring in a standardised manner
  
• Documentation performed directly from scripting
  
• Infrastructure allows one analyst to monitor 14 models (tests + documentation) in less than 10 working days
  
• Similar infrastructure being developed for other desks
  

• Solution provides effective model monitoring across asset classes, faster and at lower cost 
  
• Test results presented in well-defined presentation with comparison to pre-defined thresholds using color coding, allowing for easy analysis and enhanced decision making