Learn / Risk management

Back to learn

Should a regime filter turn a strategy off or just size it down?

Turning a strategy fully off makes sense when the regime invalidates the edge or the execution conditions behind it. Sizing down is better when the edge still exists but becomes weaker, noisier, or less deserving of scarce capital.

What to remember

  • A mean-reversion sleeve that gets run over during liquidation cascades.
  • A carry sleeve whose expected edge vanishes when crowding stress becomes extreme.
  • Any strategy whose fill quality or market depth breaks before the signal has time to matter.
  • Volatility rises, but the strategy still works with smaller bets.

Short answer

Use a hard off switch when the hostile regime breaks the actual premise of the strategy, not just its comfort level. Use a size-down rule when the strategy still has some edge but the expected payoff, liquidity, or confidence has clearly deteriorated.

The wrong mistake is pretending those are the same decision. Turning something fully off is a structural claim. Sizing down is a confidence and risk-budget claim.

When a full shutoff is cleaner

A binary gate is usually cleaner when the strategy becomes actively dangerous in the bad regime. That can happen when execution quality disappears, the sign of the edge flips, or the tail risk becomes much larger than the historical average suggests.

  • A mean-reversion sleeve that gets run over during liquidation cascades.
  • A carry sleeve whose expected edge vanishes when crowding stress becomes extreme.
  • Any strategy whose fill quality or market depth breaks before the signal has time to matter.

When sizing down is the better tool

Sizing down fits better when the edge weakens by degree instead of disappearing entirely. In that case the context layer is not saying do not trade. It is saying this is not the moment to express the same conviction with full size.

  • Volatility rises, but the strategy still works with smaller bets.
  • Several sleeves still have positive expectancy, but one should receive less capital.
  • The context signal is noisy enough that hard switches would create unnecessary churn.

How to compare the choices honestly

Run the same base strategy under both policies and compare net return, drawdown containment, turnover, and sensitivity to threshold changes. Then ask the practical question too: which rule would a human operator still trust when the market becomes chaotic and the model starts disagreeing with itself?