Volatility Regimes: Identifying and Trading Them

Defining regimes

Markets alternate between calm, trending, and crisis regimes. A simple detector: rolling 20-day realised volatility versus its own 1-year median. Crossing the median marks a regime shift. More robust variants add a second condition — for example trend strength (ADX or moving-average slope) — because volatility alone cannot distinguish a calm range-bound market from a calm strong trend.

Regime-conditional behaviour

• Correlations rise in crises — diversification erodes exactly when needed.
• Volatility clustering means high-vol periods persist.
• Trend-following strategies shine in high-vol, and mean reversion in low-vol ranges.

The correlation shift is the most consequential feature: assets that hedge each other in normal times become joint bets in a drawdown, so portfolio-level risk must be evaluated under crisis correlation assumptions, not historical averages.

Implications

Size positions by inverse volatility, and re-test strategy performance conditional on regime rather than pooled across time. Regime filters reduce drawdowns but add turnover and whipsaw risk — calibrate the filter's lag against the cost of false regime switches. A filter that flips on every volatility blip converts a stable strategy into a churning one, so hysteresis (requiring a sustained cross before switching) is usually worth the slower response.

Practical workflow

1. Compute the regime signal on a fixed frequency and store it — never retroactively reclassify history.
2. Allocate a base book sized for the calm regime and a tactical overlay for confirmed regime shifts.
3. Re-validate strategy parameters only within regime, keeping at least one out-of-sample period per regime type.

References

• Hamilton, Regime-Switching Models of the Term Structure(academic)
• Ang & Timmermann, Regime Changes and Financial Markets(academic)
• Morgan Stanley — Volatility Regimes Research(industry)