Modelling and Professionalism
Modelling and Professionalism
This domain tests the model lifecycle and the professional judgements around assumptions, validation, documentation and use of output.
Core rules and concepts
- Models simplify reality; model risk is the risk of wrong decisions from model error or misuse.
- Governance covers ownership, independent validation, change control, access and documentation.
- Assumptions need evidence, sensitivity testing and review against experience.
- Validation includes reasonableness checks, backtesting, reconciliation and challenge of expert judgement.
- Professional standards require integrity, competence, transparency and escalation of material concerns.
Worked example
A pricing model assumes 4% investment return. A candidate tests 3% and 5%, checks the model against last year's actual return, and documents that the 4% assumption is above recent experience. They run profit sensitivity and disclose the margin. If the model cannot handle negative rates, that limitation is documented and escalated rather than hidden.
Traps
Overfitting to historical data, using undocumented spreadsheet changes, presenting output without uncertainty, and treating validation as a one-off exercise. Also confusing model precision with accuracy.