Failure Modes

THE PROBLEM

VaR tells you nothing about losses beyond the confidence threshold. A 99% VaR of $10M means 1% of days will exceed $10M�but by how much? $11M? $100M? VaR doesn't say.

MITIGATIONS

• Use Expected Shortfall (CVaR) for tail risk
• Run stress tests for extreme scenarios
• Never use VaR as the only risk measure

THE PROBLEM

Correlations estimated in calm markets don't hold in crises. During stress, correlations tend toward 1.0 as everything sells off together. Your "diversified" portfolio isn't.

MITIGATIONS

• Use stress-period correlations
• Model regime-switching behavior
• Don't rely solely on correlation for diversification

THE PROBLEM

Models calibrated to historical data assume the future resembles the past. They fail when market structure changes, new products emerge, or unprecedented events occur.

MITIGATIONS

• Test on out-of-sample data
• Use walk-forward validation
• Build in model uncertainty explicitly

THE PROBLEM

Portfolio optimization assumes you can trade at current prices. In reality, large trades move markets, and liquidity evaporates exactly when you need it most.

MITIGATIONS

• Model transaction costs explicitly
• Add liquidity constraints to optimization
• Stress test with reduced liquidity

THE PROBLEM

Mean-variance optimization is highly sensitive to input estimates. Small changes in expected returns produce wildly different optimal portfolios�"error maximization."

MITIGATIONS

• Use shrinkage estimators
• Apply weight constraints
• Consider robust optimization
• Use Black-Litterman framework