A CVA number can be stable, reproducible and wrong. An engine that nets across the wrong agreements or updates collateral after default may converge precisely to the wrong cash flows. This lesson turns the valuation identity into tests of signs, limits, numerical accuracy and economic sensitivity, then sets a bounded validation assignment for the programme.
Learning objectives. By the end of this note you should be able to:
- separate implementation, calibration and model challenges;
- derive recovery and exposure sensitivities under fixed assumptions;
- design numerical convergence tests with paired uncertainty estimates;
- reconcile clean value, CVA, DVA and collateralised adjustments;
- produce a validation conclusion supported by named tests.
In this lesson
- Begin with a model contract
- Exact invariants before statistical comparisons
- A closed-form benchmark
- Sensitivities must state what stays fixed
- Numerical convergence is more than path count
- Challenge information and collateral timing
- A bounded programme validation dossier
Before you start
The full lesson, with worked examples and exercise solutions, is for members.
Read the full lessonNew to SelfMadeQuant.com? See membership options or browse the Counterparty Credit Risk and CVA syllabus.