Where to Place Your Stop Loss in ICT Trading

Nearly everyone asks this question the wrong way round. “How wide should my stop be?” assumes the answer is a distance — twenty points, thirty pips, a number you choose based on what feels survivable.

It is not a distance. It is a location, and the chart has already decided where it is before you place the order. Your only real job is to find it and not argue with it.

This page works through where that location sits in each of the common ICT setups, how far past it to sit, which swing to use when there are several, and what to do when the honest answer is more than you wanted to risk. It is the detailed companion to ICT trade management, which covers the wider picture of stops, targets and exits together.

The only rule there is

The stop goes beyond the price that would prove your reason for entering is wrong.

That is the whole thing. Every specific case below is this sentence applied to a particular setup, and if you understand the sentence you can work out placements for setups this page never mentions.

What makes it useful is that it forces a question most traders skip: what would actually have to happen for me to admit this idea was wrong? If you cannot answer that, you do not have a trade with a definable risk — you have a position with a number attached to it.

Two corollaries follow, and both are unpopular:

• The stop cannot be chosen for affordability. A price that disproves your read does not become a different price because a wider stop costs more than you want to spend.
• The stop is not a prediction of how far price will move against you. It is a statement about which prices are compatible with your idea. Price can trade anywhere; only some of those places leave your reasoning intact.

Where it goes, setup by setup

Each of these follows the same logic. Find the structural feature the entry depends on, then place the stop beyond it — never inside it, never at it.

Order block

A valid order block in this methodology is preceded by a liquidity sweep. That sweep is the feature the trade depends on, so the sweep's extreme is the invalidation, not the order block itself.

For a long: the stop goes below the low of the candle that swept sell-side liquidity, not below the order block's low. Those are often different prices, and the difference is usually where people get stopped out. If price trades back through the swept low, the sweep did not hold, and the entire premise — that buy orders were filled down there and price is now leaving — is false.

Fair value gap

For a long from a bullish fair value gap, the stop goes beyond the wick of the sweep that preceded the gap, which is the convention used throughout this site. Where there is no sweep to reference, the fallback is the gap's own far edge — for a bullish gap, the high of the first candle in the three-candle formation, which forms the gap's lower boundary.

A gap that fills completely has been rebalanced. The inefficiency that gave you a reason to expect a reaction no longer exists. Placing the stop at consequent encroachment, the gap's fifty per cent, is a common shortcut and a poor one: half the gap filling is normal behaviour the methodology explicitly anticipates, not evidence you were wrong.

Breaker block

A breaker is a level that failed in one direction and is now expected to hold in the other. The invalidation is a close back through it in the original direction — that means the breaker did not flip after all. The stop sits beyond the breaker's origin, giving room for a wick through without being closed through.

Market structure shift

If the entry is justified by a market structure shift down, the shift is invalidated by a body close back above the swing high it broke. Stop above that swing high.

The body-close requirement matters. A wick above the swing during a shift is normal; a close above it means structure has been reclaimed. If you treat a wick as invalidation you will exit good trades constantly.

Optimal trade entry

An OTE entry sits inside a retracement of a leg, and the leg's origin is the invalidation. For a long, that is the low the leg started from. Stop below it. The 79% level is not the invalidation — it is an entry zone, and using the bottom of the zone as the stop puts you inside the structure again.

Turtle soup

A turtle soup is the failed-breakout trade: price takes out an obvious high or low and reverses. The invalidation is straightforward — if price continues beyond the level it just swept and keeps going, it was a genuine break, not a raid. Stop beyond the extreme of the sweep, which is usually close to entry and one of the few setups where the honest stop happens to be tight.

How far beyond is beyond?

“Beyond the swept low” still leaves a decision: how much beyond? Too tight and you are inside the noise. Too wide and you are paying for protection you do not need.

Three principles settle it.

Use the wick, not the body. The wick is where price actually traded and where the orders were actually taken. Placing the stop at the body of the sweep candle leaves it inside the range price has already demonstrated it will visit.

Add a margin, and scale it to the day. The buffer exists to clear the exact price where stops cluster, so it has to be big enough that ordinary noise at that moment does not reach it. That is not a fixed number. On a session where NQ is swinging 200 points, a three-point margin is inside the noise; on a quiet 80-point day the same three points is generous. Judge it against the size of the recent candles rather than carrying one figure across every day.

If you have never done this and want somewhere to start, a few points past the wick on NQ and a point or two on ES is the right order of magnitude — but treat those as orientation, not a rule. The moment a buffer becomes a habit you apply without looking, it is a fixed distance wearing a different name, and this whole page is an argument against those.

Widen for spread and for news. On spot forex and gold, the spread can widen at the session open and around scheduled releases, and a stop sitting a hair beyond the level will be taken on the spread alone. If the trade is running into a high-impact release, either the stop accommodates the widening or the position should not be open through it.

Which swing is the invalidation?

Charts rarely present one obvious low. There is the low of the sweep candle, the low of the session, the low that started the leg, the low from yesterday. Choosing badly gives you either a stop that means nothing or one so wide the trade is untakeable.

The answer comes from the entry, not from the chart's appearance: the invalidation is the structural point your specific reason depends on.

If several of these apply at once, the widest of them is the real invalidation — and that is a signal about size, not a reason to pick the nearest one because it is cheaper. Swing point hierarchy covers how these levels rank against each other.

The timeframe trap

This is the most expensive mistake on the page, and it is committed by people who think they are being precise.

You identify a fifteen-minute order block. You drop to the one-minute to refine the entry, find a small structure there, and place your stop below the one-minute swing — because that is the chart you are looking at, and the stop is beautifully tight.

You have just mismatched the idea and the risk. The reason you are in the trade is a fifteen-minute level. A one-minute swing being broken tells you nothing about whether the fifteen-minute level is holding; one-minute structure breaks constantly inside a valid higher-timeframe reaction. You have taken a trade with a fifteen-minute premise and given it a one-minute invalidation, so you will be removed by noise that has nothing to do with your reasoning.

The stop belongs on the timeframe that produced the idea. Dropping down refines your entry price, and a better entry price means the same invalidation costs fewer points, which means you can take a larger position for the same risk. That is the entire benefit and it is a real one. It is not a licence to move the invalidation.

The cleanest way to hold yourself to this: mark the stop on the higher timeframe chart before dropping down. Then the number is already decided and the lower timeframe cannot talk you out of it. Top-down analysis covers the wider workflow.

Why your stop keeps getting hit

If you are stopped out repeatedly by a few ticks before the move runs without you, the explanation is usually not bad luck.

Risk management taught almost everywhere says put the stop below the recent low. That advice is followed by an enormous number of people looking at the same chart, so an enormous number of stops end up within a few ticks of the same price. A cluster of resting sell orders sitting just below an obvious low is, in this methodology, not an accident of crowd behaviour — it is precisely the kind of pool price is understood to seek out before delivering. Liquidity and the sweep cover the mechanism properly, and why price reverses right after you enter is the whole article on this experience.

The practical consequence for placement is narrow but important: being beyond the obvious level is not the same as being at it. A stop at the low is in the pool. A stop a few ticks under the low is still in the pool. A stop placed beyond the wick of the candle that already swept that pool is on the other side of the event.

There is no arrangement that makes a stop unhittable, and any material claiming otherwise is selling something. What you can do is stop volunteering.

Walkthrough — choosing between four lows

The numbers below are constructed to show the decision points cleanly. It is a teaching illustration rather than a record of a trade.

The setup. NQ, 09:58. Price has sold off through the morning and is reacting from a 15M bullish order block at 21,310–21,332. You want to be long. Four candidate lows are visible on the chart, and the one you pick decides both the validity of the trade and whether you can afford it.

Work it backwards from the reason. Why are you long? Because price swept the session low at 21,296 and reversed. That is the sentence, so 21,288 — the extreme of that sweep — is the price that disproves it. If trade returns below there, the sweep did not hold and the premise is false.

The 1M low at 21,318 is noise; it will be taken out on any ordinary retracement toward the block and tells you nothing. The order block low at 21,310 is closer than the event the trade depends on, so it sits inside the structure. And the previous day’s low is real but irrelevant here — it is not what today’s reversal was about, and using it would be paying 122 points for an idea that dies at 38.

Stop at 21,284, a few points beyond the sweep wick, scaled to a session that has been moving in 15–20 point swings. Risk is 42 points from entry.

Then the affordability question, in the right order. At 1% of a $25,000 account, the budget is $250. A 42-point NQ stop is $840, which is over three times too much. The stop does not move. What moves is the instrument: the same 42 points on MNQ is $84, so three micros is $252 — at budget, same read, same invalidation.

Had the account been $8,000, one micro at $84 would still be within 1%, and the trade remains available. Below roughly $4,000 it is not, and the correct response at that point is to skip it rather than to move the stop to 21,318 and pretend the idea died at eight points.

What that stop actually costs

The correct stop is frequently wider than people expect, and the reaction is usually to tighten it. The correct response is to change the size or the instrument. Here is what a given distance costs across the common ICT markets:

Same chart, same read, same stop distance — and a tenfold difference in what it costs to be wrong. The micros exist precisely so that the structural stop stays affordable on a smaller account, and using them is not a lesser version of the trade. It is the same trade, sized honestly.

For spot forex, a standard lot of EUR/USD is roughly $10 per pip, so a 25-pip stop is about $250; a mini lot makes it $25. On XAU/USD, a standard 100-ounce lot moves $100 per dollar of gold, which is why gold stops measured in dollars rather than cents get expensive quickly.

Run your own numbers through the position size calculator rather than estimating. Position sizing covers the method, and trading ICT with a small account works through what is realistically available at each account size.

Hard stop or mental stop?

A mental stop is a stop you intend to honour when price reaches it. The argument for one is that it keeps your order out of the visible book and out of the pool.

The argument against it is that it requires you to do the hardest thing in trading — accept a loss — at the precise moment it is hardest to do, with the position live and moving against you, and with a perfectly good reason available to wait just a little longer.

A resting stop makes that decision once, calmly, before anything is at stake. A mental stop makes it under pressure, every time, and it only has to fail once for the arithmetic of a whole month to change.

Unless you have specific evidence from your own reviewed trades that you honour mental stops without exception, use a hard stop. The small edge from hiding the order does not compensate for the occasional catastrophic override, and trading psychology explains why the override is far more likely than people believe of themselves.

Six placements that quietly cost money

The last one deserves emphasis because it is the only truly dangerous item on the list. Every other mistake costs you a normal loss. Moving a stop away from an approaching price is how normal losses become the kind that end accounts, and it is nearly always dressed up as conviction.

Frequently Asked Questions