1
Q
Model Risk
A
the potential adverse consequences from decisions based of incorrect or misused model outputs
2
Q
Finance vs Physics Modeling
A
- financial models approximate human behavior which is unpredictable
- financial parameters change frequently, making models less reliable over time
3
Q
How to manage pricing risk on actively traded products
A
- model risk is limited since market prices are observable.
- models serve as communication tools, such as when using implied volatilities or options
- interpolations between strike prices and maturities are also essential in valuation
4
Q
How to manage pricing risk on less actively traded products
A
Best practices:
- using multiple models
- calibrating them to active market instruments
- setting up reserve accounts to avoid immediate profit recognition
5
Q
Marking to Market vs Marking to Model
A
- firms use marking to market for less liquid assets (based on recent trades or broker quotes) or marking to model (using internal controls)
-> most reliable and less risk - firms use marking to model when market prices are unavailable, relying on assumptions and interpolation
-> highest risk model since the valuation heavily relies on unobservable inputs
6
Q
Model Building Pitfalls
A
Overfitting: creating a model too complex for real-world use
Overparameterization: using too many variables, making the model impractical
Complexity: Risk that users may misunderstand or misuse a complex model
7
Q
Effective Model Risk Management
A
- Essential to mitigate potential losses from incorrect or misused models
- Following regulatory guidance, validating models, and simplifying where possible can help reduce model risk
- have to keep in mind the unpredictability of human behavior and market conditions
