Model Risk

Model risk occurs when financial models produce inaccurate outputs due to flawed assumptions, incorrect data, or improper application. It's the risk that your risk models are wrong.

Sources of Model Risk:

Source	Example
Wrong Assumptions	Assuming normal distribution for stock returns
Data Quality	Using stale or incorrect input data
Overfitting	Model works on historical data but fails in new conditions
Implementation Errors	Programming bugs in trading algorithms
Misuse	Using a model outside its intended scope
Regime Change	Model trained on low-rate era fails in high-rate era

Famous Model Risk Failures:

Failure	Cause	Loss
LTCM (1998)	Models assumed stable correlations	$4.6 billion
2008 CDO Crisis	Gaussian copula model underestimated correlation	Trillions
JPMorgan "London Whale" (2012)	VaR model was manipulated	$6.2 billion
Knight Capital (2012)	Faulty algorithm deployment	$440 million

For Individual Investors:

• Retirement calculators assume historical returns continue — they may not
• Monte Carlo simulations are only as good as input assumptions
• Stock valuation models (DCF) are highly sensitive to growth rate assumptions
• Always ask: "What if my model/assumptions are wrong?"

Managing Model Risk:

• Use multiple models and compare results
• Stress test assumptions (what if growth is 50% lower?)
• Maintain healthy skepticism of precise predictions
• Include safety margins in all financial plans