Understanding Market Regimes
Strategies do not work or fail in general. They work in some conditions and fail in others, and the conditions change more often than most people account for.
Traders often search for a strategy that works. A more productive question is which conditions a strategy works in, because markets alternate between states that reward opposite behaviours.
The basic regimes
Trending
Price moves persistently in one direction. Momentum approaches thrive; mean-reversion approaches bleed steadily, because every apparent extreme keeps extending.
Ranging
Price oscillates within boundaries with no net progress. Mean reversion works; trend following generates repeated false breakouts and death by a thousand small losses.
High volatility
Large moves in both directions. Stops are hit more often regardless of direction. Position sizes calibrated for calm conditions become dangerous.
Low volatility
Compressed ranges, small moves. Breakout strategies see few signals. Notably, low volatility tends to precede high volatility — compression resolves eventually.
Identifying the current regime
- ADX — measures trend strength without direction. Readings above 25 conventionally suggest trend, below 20 suggest range.
- ATR relative to its own average — a straightforward volatility regime measure.
- Moving average configuration — when short, medium and long averages are ordered and separated, a trend is usually in place.
- Realised volatility percentile — where current volatility sits against its own recent history is more informative than the absolute number.
The unavoidable problem
Every regime measure is backward-looking. You identify the current regime using past data, which means you recognise a change after it has begun. There is no way around this, only ways to reduce the lag — each of which increases false signals.
You cannot know the regime in real time. You can know that it recently changed, which is still worth a great deal.
What to do about it
Two coherent approaches. Either specialise — run one strategy and accept flat or negative periods when conditions do not suit it — or maintain several approaches and weight them by the detected regime. The second is more robust and considerably harder, because the switching logic becomes another thing to overfit.
What does not work is optimising a single strategy until it performs well across all regimes simultaneously. That usually produces something fitted to the specific sequence of regimes in your test data.
Educational content only. Not financial advice.
See the models in action
Reading about forecasting is one thing. Create a free account and get 5 AI predictions a month — no card required.
Start free
