Strategy · Market Regime
The Same Strategy Makes Money in a Trend and Bleeds in a Range
The same parameters that ran perfectly well last year have been losing all this year. The strategy has not been touched, the code has not been touched, the server has not gone down — what changed is the market.
This is the loss most often blamed on the wrong thing. The instinct is to decide the strategy has stopped working, so back you go to retune it or swap in something new, and the result is usually worse. A more accurate description: the strategy did not stop working, the market it works in is no longer here.
What a market regime actually means
A market regime is not a difficult idea: over different stretches of time the market runs with a different character, and the same rules can behave in completely opposite ways depending on which character is in front of them.
A crude but workable split gives four of them:
| Regime | Roughly what it looks like | Comfortable in it | Having a hard time |
|---|---|---|---|
| Uptrend | Keeps working higher; pullbacks stay shallow and the move carries on | Trend following; holding and sitting still | Grid (sold out early, then left behind) |
| Sustained downtrend | Lower and lower, bounces weak and short | Sitting in cash; short-side strategies | Grid (buying all the way down); averaging into the fall |
| Range | Back and forth inside a band, both edges touched repeatedly | Grid; range strategies | Trend following (stopped out over and over) |
| Violent chop | Direction reverses often, and the swings are wide | Almost nobody | Anything carrying leverage |
No cut-off values are given here, because the threshold differs completely from one instrument and one timeframe to the next, and a number copied from somewhere else will most likely not apply. What matters is accepting one thing first: you are not facing a single market, but several that take turns coming on stage.
Grid and trend following fail in exactly opposite places
Set the two most common strategy families side by side and this becomes very easy to see.
A grid makes its money by buying a step down and selling a step up; what it eats is movement back and forth. In a range, price keeps crossing its levels and it collects a little each time. Once the market picks a direction, though — a sustained fall above all — it buys level after level on the way down, fills the whole grid, and price is still falling, with the unrealised loss compounding. On a leveraged grid, this is where liquidation happens. A one-way rally is no kinder: the position is sold out early and the entire later stretch of the move is missed; being left behind will not blow up an account, but it is quite enough to make someone chase the highs by hand.
Trend following is the mirror image. It waits for the direction to be confirmed before entering and cuts when it is wrong, covering many small losses with a few big moves. In a trending market it is comfortable. In a range, price breaks out, it enters, price turns straight back and takes the stop, then breaks out again, in again, stopped out again — no large swing to capture, and fees and stop-losses handed over each time round. That grinding is less dramatic than a blow-up, but the account curve slopes steadily downhill.
Placed next to each other it is obvious: a grid is afraid of one-way markets, trend following is afraid of ranges. The places where they fail barely overlap — good news, since in principle they complement each other, and bad news, since you must know which market you are in before you know which one to field.
Why changing strategy is usually worse than working out which market this is
The most natural move after a run of losses is to change the strategy. Changing it without first reading the regime is a bet: will the one you switch to suit whatever comes next?
Timing makes it worse. People retire a strategy only after it has lost for a while, and that stretch is often the second half of exactly the market it is unsuited to. By the time you switch the grid off, the one-way fall may be nearly done; by the time you give up on trend following, the range may be about to break. The act of switching naturally lands at the worst possible moment.
The value of reading the regime first is not precision but that the question gets asked correctly. The question is not which strategy is better; it is whether the one you already have still holds in this kind of market. If the answer is no, the sensible move is usually not a different strategy but a smaller position, or standing aside and waiting. Sitting in cash is not the absence of a strategy; it is a decision with conditions attached.
How to roughly tell which market you are in
Cold water first: there is no accurate way to tell, and certainly no way to tell in advance. Any method claiming to identify the regime precisely is worth doubting first. What is available is a rough read, and it usually arrives late.
A late read still beats no read. A few angles that need no indicators at all:
How many times the edges of the range have been touched. Pull up the recent stretch. If price has run between an upper and a lower edge many times and been pushed back each time, that looks like a range; if one side has broken cleanly and price has not come back, drop the range assumption.
Whether the original direction resumes after a pullback. The signature of a trend is that it comes back a little and then carries on. If every pullback turns into a complete journey the other way, there is no dominant direction.
The shape of volume. Where direction is clear, the pushes that go with it usually come on expanding volume, while the back-and-forth of a range tends to come on shrinking volume. The signal is crude, but read alongside price structure it offers a bit of corroboration.
Look at a bigger timeframe. What reads as a trend on a small timeframe may be half of one swing on a larger one. When judging the regime, start with a chart one or two steps above the timeframe you trade.
All of the above is a way of thinking, not a set of parameters — the moment numbers are attached to it, it turns into another strategy you have to validate yourself, and it will carry all the faults of a backtest that looks beautiful and then loses money live.
Switching strategies often costs more than you think
Once you accept that the strategy should follow the regime, the next trap arrives immediately: switching far too often.
Switching costs at least three things. The first is the cost of timing, covered above — you almost always leave where the previous strategy hurt most, turning an unrealised loss into a real one. The second is transaction cost: every switch closes the old position and builds a new one, and both legs pay fees and slippage. The third is the easiest to overlook, the cost of sample: a strategy has to run long enough before you know whether it is any good, and if you change it every few days each one leaves only a scrap of record behind. You end up permanently short of the sample needed to conclude anything.
So reading the regime is not an invitation to trade more. The sane version looks more like this: settle on a review rhythm you can keep to, and move only when the character of the market has changed clearly and has stayed changed — rather than making the call again from scratch every morning.
Rather than hunt for an all-weather strategy, write down when yours applies
For a lot of people the real turning point on the quant path is the move from looking for a strategy that makes money all the time to stating the conditions under which their strategy holds.
The first cannot be found: every strategy makes money inside some particular market structure, and structures change. The second is achievable, and the method is not complicated — write a short note on where each of your strategies applies, honestly covering three things: which markets it makes money in, which markets it loses in, and how much loss is the point at which I stop.
The benefit of writing it down is that the judgement moves out of the session and into the time before it. When things genuinely get uncomfortable, calm assessment is hard to come by, and by then the answer is already on that piece of paper. This belongs in the backtesting stage: besides returns, a backtest report should record how differently the strategy behaved across the various stretches inside the sample — for how to run that process, see How to Backtest a Strategy.
In the end, the boundary of a strategy is worth more than the strategy. Admitting that yours only works in one kind of market is not defeat; it hands the uncontrollable part back to the market and keeps only the part you can manage.
FAQ
What is a market regime?
It refers to the different ways a market runs over different stretches of time; the common crude split is trending markets, ranging markets and periods of violent movement. The same parameters can behave in opposite ways from one regime to the next: a strategy that earns from movement back and forth loses steadily in a one-way market, and one that earns from direction is stopped out repeatedly inside a range. The point of reading the regime is not prediction, it is knowing whether the strategy in your hands still holds.
When is a grid strategy most dangerous?
In a one-way market, and above all a one-way fall. A grid buys a level each time price drops a level, so a market that keeps going down means it keeps adding, and after the whole grid is filled price is still falling, with the unrealised loss widening all the way. With leverage on top, this is the liquidation scenario. A one-way rally is unfriendly too: the position is sold out early and the entire later stretch of the move is missed. Where a grid is comfortable is a range, with price crossing its levels over and over.
Why does trend following struggle so badly in a range?
Its rule is to enter once the direction is confirmed and to cut when the direction turns out to be wrong. In a range, price breaks out and turns straight back, so it enters, stops out, enters again, stops out again, paying tuition each time round. This loss is not one dramatic event like a liquidation; it grinds the account down bit by bit, and with the fees and slippage of every entry and exit on top, the curve keeps sinking. What it needs is a market with room to travel.
How do I tell whether the market is trending or ranging right now?
There is no accurate method, and certainly none that works in advance; a rough read is all that is available, and it usually lags. Things you can look at: whether price is being pushed back repeatedly between an upper and a lower edge, whether the original direction resumes after each pullback, whether volume expands on the pushes that go with the direction, and what the chart one or two steps above your trading timeframe looks like. These are ways of thinking, not parameters to be copied.
Should I switch strategies as the market changes?
Not frequently, no. Switching carries three costs: you almost always leave where the previous strategy hurt most, turning an unrealised loss into a real one; every change of position pays fees and slippage; and each strategy only runs for a short stretch, so you never accumulate enough sample to judge whether it works. Move only when the character of the market has changed clearly and has stayed changed, and set the rhythm of your reviews in advance.
Is there an all-weather strategy that suits every market?
No. Every strategy makes its money inside one particular market structure, and structures change. The more practical approach is to write down the conditions under which each strategy applies: which markets it makes money in, which markets it loses in, and how much loss is the point at which it stops. Admitting that a strategy has boundaries keeps an account alive far longer than continuing to search for one that has none.
Next time a strategy starts losing, hold off on the parameters and spend a few minutes on whether the character of the recent market has changed. If it has, the right move may be a smaller position or a pause, rather than another strategy that suits the present just as badly. Whether the strategy itself is trustworthy is a separate gate: the traps in A Beautiful Backtest That Loses Money Live are worth going through one by one before real money goes in. And to work through what quant trading is from the beginning, start with What Is Quant Trading, Really?