Volatility and Risk Notes

Volatility is both a modeling target and a practical risk signal.

In a research workflow, I care less about finding a single perfect volatility model and more about comparing forecasts under consistent assumptions. Realized volatility, historical volatility, GARCH-style models, and machine learning forecasts should be evaluated on the same data splits and loss functions.

For portfolio decisions, volatility forecasts are useful only when they connect to position sizing, drawdown control, and stress testing.