Where to Place Your Stop Loss on ICT and CRT Setups

Most traders learn where to enter and then improvise the stop. That's backwards, and it's the reason so many people take good setups and still lose money.

Here's the thing nobody tells you early enough: on smart money setups, the stop is not a risk decision. It's a chart decision. The chart tells you where the idea is wrong. Your account tells you how big to trade so that being wrong there costs an acceptable amount. Those are two separate questions, answered in that order, and swapping them is the single most expensive habit in retail trading.

This article covers where the stop actually goes on the four setups you'll trade most — liquidity sweeps, order block mitigations, fair value gap entries and breakers — how much buffer to add, and what to do when the honest stop is wider than you'd like.

The only question your stop answers

Before anything else, get the framing right. A stop loss is not "the most I'm willing to lose." It's the answer to one question:

At what price is this setup no longer the setup I took?

If your entry was based on a low being swept and price reclaiming, then a candle closing back below that low means the reclaim failed. That's the level. Not five pips, not a round number, not whatever distance makes the position size feel comfortable.

Everything that follows is just applying that question to specific setups.

And the corollary, which matters just as much: once you have that level, you do not move it closer. Tightening a stop to make a trade "affordable" doesn't reduce your risk, it converts a trade with defined invalidation into a coin flip with a smaller loss. You'll be right about direction and stopped out anyway, which is the most demoralising way to lose.

Placement by setup

Liquidity sweep (and CRT)

The setup: price runs an old high or low, takes the stops resting there, and closes back inside the range. You enter on the reversal.

The stop goes beyond the extreme of the sweep wick — the actual high or low that price printed while taking the liquidity, not the candle body and not the old level that got swept.

This is the placement people most often get wrong, and the reason is worth understanding. The old high or low you were watching was liquidity. It's gone now — price took it. The new extreme, the tip of that sweep wick, is where invalidation lives, because if price trades beyond it the "sweep and reject" story is dead and you're simply in a market that's continuing.

The same logic drives the entire CRT trading strategy: the sweep candle's extreme is the level the whole idea is built on, so it's the level your risk hangs from.

Order block mitigation

The setup: price returns to an unmitigated order block and you enter on the reaction.

The stop goes beyond the far edge of the zone, with the zone drawn from the body to the wick of the origin candle. Not the middle. Not "just past my entry."

Traders shrink this constantly, because entering at the near edge of the zone and stopping at the far edge feels like a lot of risk for a level that either works instantly or doesn't. But a stop inside the zone isn't a stop — it's a bet that price will react at the exact tick you chose within a band you yourself drew as an area. If a decisive close through the far side is what invalidates an order block, then that's where the stop goes.

If the zone is genuinely too wide to trade, the answer isn't a tighter stop. It's a lower-timeframe entry that gives you a smaller zone — see below.

Fair value gap entry

The setup: price retraces into an unfilled three-candle imbalance and continues from it.

The stop goes beyond the origin of the move that created the gap — not the far edge of the gap itself.

This one catches people out. A fair value gap filling completely is normal; gaps fill all the time and the move still works. What isn't normal is price continuing past the swing that produced the displacement in the first place. That's your level. Using the far gap edge gives you a stop that's technically tight and practically random, and you'll get clipped on ordinary rebalancing.

If you're entering at the midpoint of the gap rather than its edge — which is the standard refinement — you've already improved your reward:risk without touching the stop. That's the correct way to tighten risk: move the entry, never the invalidation.

Breaker block

The setup: an order block failed, structure shifted, and the same zone now works in the opposite direction.

The stop goes beyond the swing that created the break — the extreme price reached when it ran through the original zone.

A breaker exists because a level failed. Which means the level itself has already proven it doesn't hold, and hanging your stop off it is hanging it off something the market just discredited. The valid reference is the extreme of the move that did the breaking. Breaker blocks explained walks through how the zone flips in the first place.

How much buffer, and why "one pip below the low" gets hunted

Every one of the placements above needs a small buffer beyond the level. The question is how much, and the wrong answer is a fixed number of pips.

Think about what you're doing when you put a stop one pip below the low of a sweep candle. You are placing an order at the exact price where the market just demonstrated it goes to collect orders. You are, quite literally, becoming the liquidity for the next sweep. This isn't paranoia about brokers hunting you personally — it's that everyone who took that setup put their stop in the same obvious spot, and that cluster is a target for the same structural reason your entry was.

Two ways to size the buffer sensibly:

• Scale it to the instrument's noise. A fraction of the recent average range on your entry timeframe — say a quarter to a half of an average candle — is a reasonable starting point. Ten pips is a rounding error on NAS100 and a huge cushion on EURUSD, which is why fixed pip buffers make no sense across instruments.
• Go beyond the next minor structure, not just the level. If there's a cluster of wicks a little past your invalidation point, put the stop past the cluster. The extra distance costs you position size; being inside the cluster costs you the trade.

Whatever you choose, write it down as a rule and apply it identically every time. A buffer chosen per-trade is just a stop chosen by feel with an extra step.

When the honest stop is too wide

This is the situation that produces most bad decisions, so be clear about the options. Your setup is valid, the invalidation level is real, and the resulting risk is more than you want. You have exactly three legitimate moves:

1. Trade smaller. This is the default and it's what position sizing is for. A wider stop with a smaller position is the same money at risk as a tight stop with a big one — and only one of them survives ordinary noise. Run the numbers through the position size calculator rather than estimating.
2. Drop a timeframe for the entry. Keep the higher-timeframe idea, but find the same structure on a lower timeframe where the zone is smaller and the invalidation is closer. Your bias doesn't change; your entry precision does.
3. Skip it. If the smallest position you can trade still risks too much, that trade isn't for your account today. Passing costs you nothing. There will be another sweep.

What you may not do is move the stop closer to make the numbers work. That's option four and it's the one that quietly drains accounts.

There's a fourth check worth running before you take any of the three: does the trade still clear your minimum reward:risk once the stop is where it honestly belongs? Widening the stop shrinks your R, and a setup that was 1:3 with an imaginary stop might be 1:1.2 with a real one. The risk/reward calculator will tell you the break-even win rate that ratio demands — and if that number is above what your strategy actually produces, the correct decision is to pass, not to hope.

Moving the stop after entry

Two rules, both unpopular:

Don't move to break-even early. Moving your stop to entry the moment the trade goes green feels like risk management and usually isn't — it converts your carefully chosen invalidation into an arbitrary level that ordinary retracement will hit. Most smart-money setups retrace into the entry zone before running. If you break-even on every trade at +0.5R you will systematically remove yourself from your winners while keeping every loser. Move to break-even when structure justifies it — a new higher low above your entry on a long — not when your P&L does.

Trail behind structure, not price. If you're trailing, trail behind each new swing point the market prints, so the stop only moves when the market's own structure says the prior level no longer matters. Trailing by a fixed distance behind price is just a tighter stop applied continuously.

Stops in a backtest vs stops live

Two adjustments, because a backtested stop and a live stop are not the same thing.

Spread. On a short, your stop is triggered on the ask, not the bid you see on the chart. If your buffer is smaller than the typical spread on your instrument, your backtest will show trades surviving that would have stopped out live. Widen the buffer or accept that your results are optimistic.

Gaps. A stop is not a guarantee of price, only of exit. Weekend gaps and news releases can fill you well beyond your level, which is another argument for sizing so that a single trade going badly wrong is survivable rather than defining.

The only way to know whether your stop placement rule actually works is to run it over a real sample — the same setup, the same buffer rule, a few hundred times. Our sister site handles that side: how to backtest a strategy without fooling yourself covers the ways a stop rule can look great on a chart you've already seen and fall apart on one you haven't.

The mistakes worth naming

• Sizing first, stop second. Deciding you'll trade one lot and then finding a stop that fits is the root error. Every other mistake here descends from it.
• Round numbers. Stops at 1.0850 or 20,000 sit exactly where everyone else's do. Use structure, not psychology.
• Same stop distance on every instrument. Gold and EURUSD do not have the same noise. A fixed pip stop is a different trade on each.
• No stop, "I'll watch it." You will not. And the one time you step away is the one that matters.
• Widening a stop that's already been placed. The mirror image of tightening, and worse. If the level's hit, the idea was wrong — that's information, not an inconvenience.

Where this fits in the bigger picture

Stop placement is the last step of a sequence, and it only works if the steps before it were done properly. You need the liquidity pool identified, the structure shift confirmed, and the zone marked — and then the invalidation level falls out of that work almost automatically. If you're ever unsure where the stop goes, it's usually a sign the setup itself wasn't clear.

If those earlier steps are still fuzzy, start with ICT concepts explained for beginners and the CRT mitigation strategy, which is the specific combination this stop logic was written for.

And when you want the whole thing taught as one system rather than assembled from articles — the sweep, the structure shift, the mitigation entry, and exactly where risk sits on each — that's what the CRTLAB ICT and CRT × Mitigation course is. One payment, lifetime access, and the execution detail that most free content leaves out precisely because it's the part that's hard to teach.

FAQ

Where exactly do you put a stop loss on a liquidity sweep? Beyond the extreme of the sweep wick — the actual high or low price printed while taking the liquidity — plus a buffer scaled to the instrument's noise. Not at the old level that was swept, because that liquidity has already been taken and the level no longer means anything.

Should my stop go above the order block or inside it? Beyond the far edge of the zone. A stop inside the zone is a bet on price reacting at one specific tick within an area you drew as a range. If a decisive close through the far side is what invalidates the order block, that's where invalidation belongs.

How many pips should my stop loss be? There is no universal number, and any answer given in pips is wrong across instruments. The stop distance is whatever the chart's invalidation level dictates; your position size then adjusts so that distance costs an acceptable amount. Ten pips is noise on an index and a wide stop on a major FX pair.

What do I do if the stop is too wide for my account? Trade smaller, drop a timeframe to find a tighter version of the same setup, or skip the trade. Those are the only three valid options. Moving the stop closer to fit your preferred size removes the invalidation that made the setup tradeable in the first place.

Should I move my stop to break-even? Not automatically, and not on a P&L trigger. Most smart-money setups retrace into the entry zone before expanding, so a mechanical break-even at +0.5R removes you from winners while keeping every loser. Move it when structure justifies it — a new swing point forming beyond your entry.

Why do I keep getting stopped out right before price goes my way? Almost always because the stop sat in the obvious place: one pip past the wick, or on a round number, where every other trader's stop sat too. That cluster is itself a liquidity pool. Adding a buffer proportional to the instrument's recent range fixes most of it.

Does stop placement change between forex, indices and crypto? The logic doesn't — invalidation is invalidation. The buffer does, substantially, because the instruments have completely different noise profiles. Scale the buffer to a fraction of the recent average candle range on your entry timeframe rather than using a fixed distance.

Is a wider stop with smaller size really the same risk? In cash terms, yes — that's arithmetic, and it's exactly what a position size calculator does. What differs is survivability: the wider stop sits outside ordinary noise, so it's hit when you're actually wrong rather than when the market wobbles. Same risk, far better hit rate.