The strategy that made you money last month is not broken. It just stopped matching the market. That gap is where most of the damage happens, and it usually happens quietly. A trend-following system keeps buying breakouts into a range that has started chopping, and every breakout fails by a few ticks. You do not notice the regime changed until the equity curve tells you, and by then you have paid for the lesson three or four times.
The fix is boring, which is why people skip it. You decide in advance which rules run in which market phase, you write the transition conditions down before you need them, and then you follow the document instead of your gut when things get loud. I want to walk through how I actually structure that, because the idea is easy and the discipline is not.
Name the phases before you name the rules
You do not need a fancy classifier. Most of the edge comes from being honest about which of three states you are in. Trending, where price is making higher highs or lower lows and pullbacks get bought or sold quickly. Ranging, where price keeps returning to a middle and the edges hold. And high-volatility, where the range and the trend both get overwhelmed by size and speed, gaps widen, and stops that felt safe get run.
Pick two or three signals to define each state and keep them dumb. A long moving average slope for trend direction. The ratio of average true range to price for volatility. Whether price is inside or outside a recent channel for range versus breakout. The point is not precision. The point is that you and a colleague looking at the same chart would agree on the label most of the time. If your regime definition is so subtle that reasonable people disagree, it will not survive a stressful Tuesday.
Write the definition as a rule you could check in ten seconds. Something like: trending if the slope is positive and price is above the channel midline; ranging if volatility is below its median and price is inside the channel; high-volatility if ATR-to-price is in the top quartile of its recent history. Historically, markets spend a lot more time ranging than trending, so do not build a playbook that quietly assumes a trend is always coming.
Map each regime to specific actions
Now the part that does the work. For each phase, you pre-commit to a strategy family, a sizing rule, and a stop style. Not a vibe. A specific set of actions you could hand to someone else.
- Trending. Run breakout or pullback continuation. Size normally, and let winners run with a trailing stop set off a moving average or a chandelier-style ATR trail. Take partials into strength if that suits you, but the default is you stay in the direction of the trend until the trail gets hit.
- Ranging. Switch to mean reversion. Fade the edges of the channel toward the middle. Size the same or slightly smaller, and use tight, fixed stops just beyond the range boundary, because the whole thesis dies the moment the edge breaks. Targets are the opposite edge or the midline, not the moon.
- High-volatility. This is the phase where the honest move is to do less. Cut size by a third or a half. Widen stops in price terms so normal noise does not knock you out, but keep the dollar risk per trade constant by trading fewer units. Prefer strategies that profit from the volatility itself rather than picking a direction. If nothing fits, cash is a position.
The reason to write this as a table you can read at a glance is that when the regime actually flips, you will be stressed, the screen will be moving, and you will be tempted to negotiate with yourself. A document that already made the decision removes the negotiation.
Write the transition rules, not just the states
Defining the phases is the easy half. The expensive half is the boundary between them, because that is where whipsaw lives. If your rule is "switch playbooks the instant the label changes," you will flip back and forth every time price wobbles across a threshold, and you will run the ranging playbook for one bar, then the trending one for the next, and lose money in both.
Two things help. First, add hysteresis. Require the new regime to persist for a set number of bars before you act on it, and make the condition to leave a regime stricter than the condition to enter the one you are already in. If you are in a trend, demand more evidence to call it dead than you demanded to call it alive. This is the same trick a thermostat uses so it does not click on and off every thirty seconds.
Second, define a neutral zone. When the signals disagree, when the slope is flat and volatility is middling, you are not in any regime, and the correct action is to trade smaller or not at all rather than force a label. Most bad regime-switching happens in this mushy middle where someone insists on being in a phase that the market has not committed to.
There is also a mechanical cost to switching. Every time you flip playbooks you pay spread, fees, and the chance of getting caught wrong-footed on the change itself. Bake that into the decision. If two adjacent regimes would trade almost the same anyway, do not bother switching. Reserve the transition for cases where the rules genuinely diverge.
A checklist you can run in a minute
Here is the routine I actually use before a session, and again any time the market feels different from an hour ago.
- Read the three regime signals and write down a single label. If they disagree, the label is neutral.
- Confirm the label has held for the minimum number of bars. If it just changed, wait unless the move is violent enough to override.
- Pull up the row for that regime: strategy family, size, stop style. Set your orders to match before you look for a trade.
- Check that your current open positions belong to the current regime. A mean-reversion short you opened in a range does not automatically become valid when the market starts trending against it. Manage legacy positions by the regime you entered them in, and be quicker to close them.
- Note the condition that would flip the regime, so you know in advance what you are watching for and are not surprised by it.
The most common failure I see is not picking the wrong strategy. It is running the right strategy one regime too long, holding the trend-following posture deep into a range because admitting the trend is over feels like giving up on a winner. The checklist exists mostly to force that admission on schedule instead of letting ego set the timing.
None of this requires exotic tools. You can keep the whole thing in a one-page document and a couple of indicators. If you want to pressure-test which definitions and stop styles actually held up across different phases, that is what backtesting and a market scorecard are for, and it is a big part of why we built the Labs and scorecard tooling into Blockcircle. But the playbook itself is just a page you write once, argue with, and then have the discipline to obey when the phase changes and you would rather improvise.