LEARN · EN · easyquanttrading.com
Why strategies stop working: regime changes
Almost every strategy has a period where it worked beautifully and a period where it did nothing. Sometimes that is decay, and sometimes the market simply changed character. Telling the two apart is difficult, and the cost of guessing wrong runs in both directions.
- Regimes differ in volatility, correlation and trend persistence
- A strategy tuned to one regime behaves differently in the next
- Detecting the change after the fact is easy; in real time it is not
- The realistic responses are diversification and smaller size, not prediction
What a regime actually is
A regime is a period during which the statistical character of the market is roughly stable: how much it moves, how correlated different instruments are, and whether moves tend to continue or reverse. Nothing announces the boundary; it is identified in hindsight.
The practical point is that the same rule can be profitable in one regime and unprofitable in the next without anything being wrong with the rule. Trend-following systems do well when moves persist and badly in choppy ranges. Mean-reversion systems do the opposite. Neither is broken when it fails; it is simply in the wrong regime.
The signals that something has changed
None of these is a reliable timing signal, and all of them are useful context when a strategy's behaviour departs from its history.
| Observation | What it suggests |
|---|---|
| Volatility well outside its historical range | Position sizes calibrated to old volatility are now wrong |
| Correlations rising across assets | Diversification is weaker than it appeared |
| Trade frequency far from the backtest | The entry conditions are firing in a different environment |
| Average win/loss ratio shifting | The payoff structure the strategy relied on has changed |
| Drawdown duration exceeding the historical worst | Worth attention, though not automatically a signal to stop |
Context, not signals. Each one changes the weight you give your prior, not the decision itself.
Why real-time detection is so hard
Every strategy has losing periods. A regime change begins as one of those, and there is no test that distinguishes them on the first day. By the time the evidence is clear, the drawdown has already happened.
This is why attempts to build regime detectors usually end as another overfitted model: the detector is tuned on historical regime boundaries that were themselves identified with full knowledge of what came next.
The responses that actually work
- Diversify across strategy types, so that when one regime hurts trend following it helps reversion.
- Size positions from current volatility rather than a fixed lot, so the account adapts even when the strategy does not.
- Set a pre-defined rule for when to reduce or stop, decided in advance rather than during the drawdown.
- Avoid the temptation to re-optimise during the losing period; that is how a temporary drawdown becomes a permanent loss.
- Accept that you will sometimes stop a strategy just before it recovers, and that this is the cost of having a rule at all.
The honest position
You cannot predict regime changes and you should be sceptical of anyone who claims to. What you can do is avoid depending on a single regime, size so that being wrong is survivable, and decide your stopping rule in advance so that the decision is not made by the version of you that is losing money.
FAQ
- How do I know if my strategy has decayed or the regime changed?
- In real time you often cannot. A useful distinction afterwards: if the logic still matches how the market behaves but returns fell, the edge may have been arbitraged away; if the market's volatility or correlation structure changed, it is a regime effect.
- Should I stop trading a strategy during a bad period?
- Only according to a rule you set in advance. Stopping during a drawdown because it feels bad is how traders convert a recoverable drawdown into a locked-in loss, and then re-enter at the worst possible moment.
- Do regime changes affect all strategies equally?
- No, and that is the point of diversification. A volatility regime that hurts trend following often helps mean reversion, and vice versa. Running both is the cheapest structural protection available.
More guides
- How EasyQuant validates strategies — evidence you can filter
- Honest backtesting, not pretty curves
- Gold strategy research that stays honest
- Overfitting detection: catch it before you deploy
- System Forge: design, then prove
- Walk-forward analysis: the only backtest that fights overfitting
- MT5 export without custody
- Glass box, not black box AI signals
Not investment advice. Historical results do not guarantee future performance. EasyQuant is a research factory — you execute on accounts you control.