Is stop hunting real? What our gold data shows

Short answer: price does sometimes run through a crowd of resting stops and turn, and that is visible on any chart. What our data does not support is the idea that the obvious levels get hit more often than any other price the same distance away, or that most stop-outs need an explanation beyond ordinary movement. On two years of hourly gold data, a stop one average true range (ATR) away was touched within 24 hours 73% of the time, whichever way the trade pointed.

That is a statement about the data we have, not about every broker: we hold no broker data, so we cannot say what any particular broker does with its clients' orders. Below is what we measured, how to tell a sweep when you see one, and what actually lowers the odds of being stopped out.

See it on a real chart: Gold (XAUUSD) liquidation heatmap — the modeled clusters above and below the current price.

What a stop hunt means

A stop hunt is the idea that price is pushed to where stop orders rest - just under a swing low, just above a swing high, past a round number - so that those orders fire, after which price comes back. A stop is a resting order that becomes a market order when it triggers, so a cluster of them going off together adds a burst of selling below price or buying above it. Someone who needs to fill a large order can use that burst as liquidity. No conspiracy is needed for that; it follows from where the orders sit.

The stronger claim is the personal one: that someone sees your stop and goes after it. We cannot test that. What we can test is whether the levels where stops are supposed to crowd behave differently from any other price at the same distance.

Stop hunt vs liquidity sweep vs liquidity grab

• Stop hunt: price trades through a level where stops are thought to rest, triggers them and comes back. The name implies that someone did it on purpose.
• Liquidity sweep: the same picture without the intent. Price takes out a prior high or low - the 'liquidity' of the stops beyond it - and closes back inside the range. It is the term smart-money-concepts traders use.
• Liquidity grab: mostly used for the same thing, often for a quicker, single-wick version. There is no official definition that separates it from a sweep; educators draw the line in different places.

All three describe one chart shape: a wick through an obvious level and a close back on the other side. The name says nothing about whether the move was planned, and nothing about what price does next.

How often an ordinary stop gets hit on gold

Before calling anything a hunt, it helps to know how often noise alone takes a stop. We treated every hourly close on gold from June 2024 to October 2026 (13,774 candles) as an entry, long and short, and checked whether a stop at each distance was touched within the next 4 and 24 hours. Distances are in ATR(14) on the 1-hour chart, so they scale with how busy the market is.

• Half an ATR: hit within 4 hours 66% of the time, within 24 hours 86%.
• 1 ATR: 40% within 4 hours, 73% within 24 hours.
• 1.5 ATR: 23% and 60%.
• 2 ATR: 13% and 50%.
• 3 ATR: 5% and 33%.

On 29 October 2026 one hourly ATR on gold was about $15.50 - roughly 155 pips, counting $0.10 as a pip as most retail XAUUSD quotes do. A stop that close was taken out within a day nearly three times in four with nobody needing to hunt it. Only at about two ATR did a stop survive the next 24 hours half the time. A trader whose stops sit inside one ATR will be stopped out on most days, and plenty of those stop-outs will look as if price came for them.

The same table priced at today's volatility, in dollars and pips, is on the gold stop-loss data page. The rates are recomputed weekly.

When you enter matters as much as where

The same 1 ATR stop was not equally likely to be hit at every hour. Opened around 13:00 London time, in the run-up to the New York open and the US data releases, it was touched within 4 hours 61% of the time, and around 12:00, 59%. Opened between 18:00 and 20:00 London time it was touched 22-23% of the time. Single-hour figures move by several points when measured on later data; the busy-versus-calm pattern held.

Scheduled US data stretch the noise further. Since January 2025, gold's range in the hour of the US jobs report (NFP) had a median of 2.3 times the same hour on ordinary days (19 releases), in the hour of a Fed decision 4.2 times (15 decisions) and in the CPI hour 1.4 times (21 releases). A wick through a stop in that hour is what a release hour normally looks like.

Every release since January 2025, with the next date and time, is on the gold NFP-day page, which links to the CPI and Fed-decision pages.

Are the obvious levels hit more often? Our placebo test

If stops were being hunted where they crowd, those levels should be reached more often than ordinary prices the same distance away. That is testable. We took the stop bands our model draws - its estimate, from volatility and recent price action, of where the crowd's stops concentrate - and paired each with a placebo: a level the same distance from price, mirrored to the other side. Then we did the same for the levels traders call obvious: previous-day highs and lows, swing highs and lows, and round numbers.

Across 28,948 modeled stop bands on our markets, price reached the band within 24 hours 47.4% of the time, and the placebo 48.1%. Where price did arrive, it reversed at 57.0% of 13,666 touched bands, against 56.8% at the placebo. Previous-day highs and lows, 20-bar swing highs and lows and round numbers came out between 1.7 and 0.5 percentage points below their placebos.

In plain terms: the places where stops are supposed to sit were not reached more often than anywhere else at that distance, and price did not turn there more often. The method and every figure are on the methodology page.

That does not prove that nobody ever runs stops. It means that, on the data we can see, sitting on the obvious level did not make a stop more likely to be hit than sitting at the same distance anywhere else. Distance and the hour did.

How to identify a stop hunt, and what it does not tell you

After the fact, a sweep is easy to spot:

• A level many traders can see: a clear prior high or low, a round number, or a shelf of equal highs or equal lows.
• A candle that trades through it, usually with a long wick, and closes back on the original side.
• Often a burst of range at that moment - on gold, often in a busy hour or around a data release.

What it does not tell you is what comes next. In our test of 5,243 sweeps of modeled stop clusters on bitcoin, price reversed afterwards only 48% of the time. A wick through a level followed by a reversal is memorable; the same wick followed by a continuation is called a breakout, and it happened about as often. Read a sweep as a description of what price did, not as an entry signal.

How to avoid stop loss hunting

If most stop-outs are noise rather than targeting, the fixes are about distance, timing and size:

• Set the stop by volatility, not by your balance. On gold a stop inside one hourly ATR was hit within a day about three times in four. Read the current ATR on the timeframe you trade before you choose the distance.
• Mind the hour. The same stop opened around midday London time was hit within 4 hours more than twice as often as one opened in the London evening. If you hold through a jobs report, CPI or a Fed decision, expect that hour to be well wider than usual.
• Size the position to the stop. A wider stop only works if the position shrinks so the loss at the stop stays what you planned. If the sensible stop leaves a position too small to bother with, skipping the trade is a valid result.
• Add your broker's spread. The stop fires on the bid or ask, and spreads widen at the session opens, the daily rollover and around releases - which is how a stop is hit by a wick the chart never showed.
• Moving the stop a few pips past the obvious level helps only by the distance it adds: in our tests a stop on the level was hit no more often than one the same distance away.

The sizing step is arithmetic: the free gold position size calculator turns your balance, risk and stop into lots, and shows the stop in multiples of the current hourly ATR.

On a prop firm challenge the daily loss limit adds a second limit: the prop firm gold stop calculator shows how many full stops of a given size fit in the day and how often a stop that far was hit within 4 and 24 hours.

So, is stop hunting real?

Stops are real liquidity, and price does sometimes run through a crowd of them and turn. But on two years of gold data, distance and timing explain most stop-outs: stops inside normal hourly movement, opened in the busiest hours or held through a release. The levels where stops are supposed to crowd were not reached, or reversed at, more often than a placebo. If your stops keep getting hit, the usual fix is a wider stop with a smaller position, not a better hiding place.

The same question from the trader's side - spreads, sizing and the order of decisions - is in why price keeps hitting your stop loss.