Regime Detection in Strategy Allocation

Markets move between distinct regimes: low volatility trending periods, range-bound consolidation, and high volatility crisis episodes. A strategy backtested across all regimes produces an average performance number that may not represent any actual trading environment. Mean-reversion strategies tend to work in ranging markets but fail in strong trends. Momentum strategies do the opposite.

We classify regimes using realized volatility ratios and VIX term structure. When short-term realized volatility exceeds long-term realized volatility by a threshold, the market is in a high-volatility state. When the VIX futures curve inverts (backwardation), it signals acute stress. These classifications are computed from observable data with no forward-looking information.

Position sizes are scaled by regime. In low-volatility environments, strategies run at full allocation. In high-volatility regimes, tactical positions are reduced by 50%. In crisis regimes, tactical exposure drops to zero and only the core allocation remains. This scaling is automatic and rule-based. It does not require a judgment call during a drawdown.

We test each strategy separately by regime using conditional walk-forward validation. A strategy must show positive out-of-sample performance in at least two of three regime types to qualify. A strategy that only works in low volatility is fragile. A strategy that works across regimes has a more durable edge.

References

• Ang, A. and Timmermann, A. (2012). "Regime Changes and Financial Markets." Annual Review of Financial Economics.
• Hamilton, J. D. (1989). "A New Approach to the Economic Analysis of Nonstationary Time Series and the Business Cycle." Econometrica.
• Kritzman, M., Page, S., and Turkington, D. (2012). "Regime Shifts: Implications for Dynamic Strategies." Financial Analysts Journal.