Why does stress testing not replace VaR for an option-heavy portfolio?

Because the two tools answer different governance questions.

Stress testing tells you what happens under selected severe scenarios. That is excellent for exposing vulnerability to curve shocks, volatility jumps, or liquidity breaks. But it does not tell you where those scenarios sit inside the full loss distribution or how losses compare across routine versus tail days.

VaR or expected shortfall still matters because they:

• summarize risk at a chosen confidence level
• support limit frameworks and regular monitoring
• allow more consistent comparison across desks and time periods

For an option-heavy book, a strong setup is usually:

• Monte Carlo or historical simulation for the distribution-based measure
• stress tests for named severe scenarios

That way you capture both tail frequency logic and scenario-specific vulnerability.