Trading concepts

Stop Hunting: What a Stop Run Is, What Real Bars Show and How to Place Stops

The theory says large traders push price through levels where stops cluster, then reverse. What a stop hunt is, how often breaks of a 20-bar low closed back above it on four real series compared with shuffled bars, and how often stops placed beyond the low were still reached.

· Founder of PatternsWizard
Checked 9 min read
Donchian Channel(20) on Bitcoin, daily, August 30, 2026 to October 28, 2026. Illustration of how the indicator reads, not a backtest.
On this page
  1. What is stop hunting?
  2. How we measured stop runs
  3. Does price bounce after a sweep?
  4. How to avoid stop hunting
  5. Do brokers hunt stop losses?
  6. Frequently asked questions

Stop hunting is the idea that large traders push the price through a level where many stop orders are likely to sit, such as just under a recent low, to trigger those stops and then trade the other way. The move it describes is real and easy to see on a chart: price breaks a low, the stops fire, and the bar closes back above. Whether anyone planned it is something no chart can show. We counted how often such breaks closed back inside on four real series, compared the counts with randomly shuffled bars, and measured how often a stop placed beyond the level was still reached in the bars that followed.

Donchian Channel(20) on Bitcoin, daily, June 1, 2026 to October 28, 2026. Illustration of how the indicator reads, not a backtest.
  • Price, daily candles (hollow up, filled down)
  • Upper, highest high of the previous 20 bars (solid line, second color)
  • Lower, lowest low of the previous 20 bars (solid line)
  • Middle, (upper + lower) / 2 (dashed line)
  • Close above the upper line (new 20-bar closing breakout, first day)
  • Close below the lower line (first day)
  • Shaded area between the bands
Donchian Channel(20) on Bitcoin, daily, June 1, 2026 to October 28, 2026. Illustration of how the indicator reads, not a backtest.Daily Bitcoin with the lowest low and highest high of the previous 20 bars. Bars whose wick crosses the lower line and close back above it are the sweeps counted below, such as June 24, June 25 and July 1, 2026.

What is stop hunting?

The theory starts from where stops are often placed: below the last swing low for a long, above the last swing high for a short, or at a round number. A sell stop becomes a market order when it triggers, so a cluster of them adds a burst of selling at the level. If the price is pushed there, that burst gives a large buyer shares to buy and can carry the price a little further before it turns. Stop hunting, also called a stop run or a liquidity sweep, describes that sequence. How a stop turns into a market order is explained on stop loss order.

Why a stop where everyone else has one gets hit

Why a stop where everyone else has one gets hit: a price path above an obvious support level. Many stop orders sit just under that level. A quick dip below the level takes out every stop, then price recovers and climbs without the traders who were stopped out.
Schematic, not market data.

Two points matter before acting on the idea. First, the same picture appears without any hunter: a low that is tested and holds, then a rebound, is ordinary price action. Second, the chart shows the move, not the orders behind it, so a "stop hunt" is labeled after the fact. False breaks in the other direction are covered on bull trap.

How we measured stop runs

Rule: a break is a bar whose low goes under the lowest low of the previous 20 bars, the lower line of the Bitcoin chart at the top of the page; a sweep is a break that still closes above that level. We counted breaks and sweeps on Bitcoin daily (March 8, 2025 to October 28, 2026), the S&P 500 ETF daily (January 4, 2021 to October 28, 2026, dividend-adjusted), Bitcoin hourly (October 4 to 28, 2026) and EUR/USD daily (March 15, 2021 to June 30, 2023). We then ran the same counts on 1,000 shuffled versions of each series, in which every bar keeps its open, high, low and close as ratios to the previous close but the order of the bars is random, so nobody can be hunting anything. The shuffle is a Fisher-Yates shuffle driven by the mulberry32 generator with a fixed seed (20261102), one generator for the 1,000 shuffles of each series, and the middle 95% runs from the 2.5th to the 97.5th percentile of the 1,000 results, with linear interpolation, so the same numbers come out each time.

Breaks of a 20-bar low that closed back above it

Breaks of a 20-bar low that closed back above it. Bitcoin daily (55 breaks): Real bars 56.4%, Shuffled bars, median 53.7%; S&P 500 ETF daily (124 breaks): Real bars 47.6%, Shuffled bars, median 46.8%; Bitcoin hourly (61 breaks): Real bars 49.2%, Shuffled bars, median 52.3%; EUR/USD daily (103 breaks): Real bars 54.4%, Shuffled bars, median 46.5%. Our computation; the inputs and the method are in the text.
  • Real bars
  • Shuffled bars, median
Our computation; the inputs and the method are in the text.Breaks of the lowest low of the previous 20 bars. Real series against 1,000 shuffled versions of the same bars. Description of four samples, not a trading test.

About half of the breaks of a 20-bar low closed back above it: 56.4% of 55 breaks on Bitcoin daily, 47.6% of 124 on the ETF, 49.2% of 61 on Bitcoin hourly and 54.4% of 103 on EUR/USD. Every real figure sits inside the middle 95% of its shuffled series (41.8% to 64.9%, 37.7% to 54.8%, 41.2% to 62.7% and 37.1% to 56.0%). Breaks that reverse within the bar are what random price paths produce at about this rate, so the count alone is no evidence of hunting.

Does price bounce after a sweep?

If a sweep marked the end of a stop run, the price should rise more often than usual afterwards. The chart compares the share of sweeps followed by a higher close five bars later with the same share over all bars: 30 sweeps on Bitcoin daily, 58 on the ETF, 30 on Bitcoin hourly and 56 on EUR/USD, the sweeps of the last five bars of each series being left out.

Share of higher closes five bars later: after a sweep and over all bars

Share of higher closes five bars later: after a sweep and over all bars. Bitcoin daily (30 sweeps): After a sweep 33.3%, All bars 49.7%; S&P 500 ETF daily (58 sweeps): After a sweep 56.9%, All bars 59.6%; Bitcoin hourly (30 sweeps): After a sweep 36.7%, All bars 48.0%; EUR/USD daily (56 sweeps): After a sweep 50.0%, All bars 47.0%. Our computation; the inputs and the method are in the text.
  • After a sweep
  • All bars
Our computation; the inputs and the method are in the text.A sweep is a break of the 20-bar low that closes back above it. Description of four samples, not a trading test.

With 30 to 58 cases, these gaps are within sampling noise: on the shuffled series, the middle 95% of the same share after a sweep spans 28 to 39 percentage points (30.8% to 69.7% on Bitcoin daily), and all four real figures fall inside it. On both Bitcoin series the gap goes the wrong way for the theory. A swept low is not, on these data, a reliable buy signal.

How to avoid stop hunting

A stop stays in place after the break, so the test is whether the price reached it at any point in the following bars, not only on the breaking bar.

Stops beyond a broken low, and the five bars that follow

Stops beyond a broken low, and the five bars that follow: candles fall to a recent low, one bar breaks under it and closes back above. Three stop levels sit under the broken low: just under it, half an average true range beyond it and one average true range beyond it. The breaking bar hits the stop just under the low; a bar two sessions later hits the stop half an ATR beyond; the stop one ATR beyond is not reached, and the price then rises.
ATR
average true range
Schematic, not market data. One possible path; on the page, how often each stop was reached is counted on real bars.

Rule: for each break, take the 14-bar average true range (Wilder's smoothing) as of the bar before the break, and check whether the low went at least 0.5 or 1 ATR beyond the level on the breaking bar or any of the next 5 bars. Breaks in the last 5 bars of each series are left out, which leaves 54, 123, 61 and 103 breaks. In the chart below, the second bar of each pair is the median of the same count on the 1,000 shuffled versions.

A buffer cuts the hits on the breaking bar itself: a stop 1 ATR beyond the level was crossed on that bar on 12% to 23% of the breaks. But over the next five bars it was still reached on about half to seven in ten of the breaks (51% to 71%), and a stop 0.5 ATR beyond on three quarters or more (76% to 85%).

Stops beyond the broken low reached within five bars

Stops beyond the broken low reached within five bars. Bitcoin daily, 0.5 ATR: Real bars 75.9%, Shuffled bars, median 74.2%; Bitcoin daily, 1 ATR: Real bars 59.3%, Shuffled bars, median 50.0%; S&P 500 ETF daily, 0.5 ATR: Real bars 80.5%, Shuffled bars, median 75.8%; S&P 500 ETF daily, 1 ATR: Real bars 62.6%, Shuffled bars, median 54.6%; Bitcoin hourly, 0.5 ATR: Real bars 85.2%, Shuffled bars, median 77.1%; Bitcoin hourly, 1 ATR: Real bars 70.5%, Shuffled bars, median 56.4%; EUR/USD daily, 0.5 ATR: Real bars 77.7%, Shuffled bars, median 76.2%; EUR/USD daily, 1 ATR: Real bars 50.5%, Shuffled bars, median 54.4%. Our computation; the inputs and the method are in the text.
  • Real bars
  • Shuffled bars, median
Our computation; the inputs and the method are in the text.Share of breaks where the low went at least 0.5 or 1 average true range under the broken level, on the breaking bar or one of the next five. 54, 123, 61 and 103 breaks.

All eight five-bar figures lie inside the middle 95% of their shuffled series, but seven are above the shuffled median. At 1 ATR the gap is 8 to 14 points on three series, each near the top of its range: 59.3% against 50.0% on Bitcoin daily (top of the range 65.4%), 62.6% against 54.6% on the ETF (64.9%) and 70.5% against 56.4% on Bitcoin hourly (70.8%). Shuffling removes the clustering of volatility, and real breaks come in volatile stretches: at the median break, the average true range of the breaking bar and the next five was 1.16, 1.22 and 1.26 times the ATR before the break on those three series, against 0.98, 1.01 and 1.06 on the shuffled ones (median of the 1,000 shuffles). EUR/USD, at 1.06 against 1.01, is the one series where the real figure at 1 ATR is below the shuffled median. Nothing here points to a hunter, but a far stop was reached more often on real bars than on random ones: a buffer buys room, not safety. Each loss is also larger, so for the same risk in money a stop twice as far away means half the size. Practical rules:

  • Place the stop where the trade idea is wrong, not at the first obvious level, with a buffer in ATR units, which scale with the market (see average true range), rather than a fixed number of points.
  • Size the position from the stop distance, as explained in our stop placement guide.
  • Avoid stops at round numbers and exactly at the prior high or low, the places the theory says stops gather.
  • Accept that no placement avoids every run.

Do brokers hunt stop losses?

When a dealer fills your stop, and where NFA Rule 2-43 comes in

When a dealer fills your stop, and where NFA Rule 2-43 comes in: three steps from top to bottom. One, the stop is triggered on the dealer's own quote and becomes a market order. Two, the order is filled at the next price the dealer offers, which can be worse than the trigger, with the dealer as the other side of the trade. Three, after the fill, NFA Compliance Rule 2-43 bars the dealer from changing the executed price, except in narrow cases such as an adjustment in the customer's favor. A shaded box at the bottom gives the check left to the trader: compare the fill with another quote from the same moment.
Steps 1 and 2 are how a dealer fill works; step 3 is NFA Compliance Rule 2-43, simplified (it allows a few narrow exceptions, such as a change in the customer's favor when settling a complaint), which applies to forex dealer members of the NFA.

A dealer that takes the other side of your trade gains when you lose, a conflict of interest that rules limit after the fact. For forex dealers that are members of the NFA, the US futures self-regulator, Compliance Rule 2-43 says a dealer "may not cancel an executed customer order or adjust a customer account in a manner that would have the direct or indirect effect of changing the price of an executed order," except in narrow cases such as settling a customer complaint in the customer's favor. Quotes that trigger stops still differ between dealers, so check another source when a stop fills at a price you did not see elsewhere. Retail traders and other participants are compared on retail trader. When the dealer keeps the other side, the trade is zero-sum between client and dealer before costs, as our page on the zero sum game explains.

More definitions are in trading concepts.

Every pattern's results on one sheet

The only pattern cheat sheet with tested stats: 152 patterns, what the textbook says and what really happened on 6 markets. Free PDF, sent to your inbox.

The PDF by email, plus our newsletter: new tests and results that change. Unsubscribe anytime. Privacy.

Frequently asked questions

What is stop hunting in trading?

A move that pushes the price through a level where many stop orders sit, such as just below a recent low, so that the stops trigger, after which the price turns back. Whether a given move was planned cannot be seen on a chart.

Is stop hunting real?

The pattern is real, but on our four series breaks of a 20-bar low closed back above it about as often as in 1,000 randomly shuffled versions of the same bars, so the pattern alone does not prove that anyone hunted the stops.

How do you avoid stop hunting?

No placement avoids it. Putting the stop beyond the obvious level by a buffer in ATR units cuts the hits on the breaking bar, but in our test a stop 1 ATR beyond a broken 20-bar low was still reached within five bars on 51% to 71% of the breaks. Place the stop where the trade idea is wrong and size the position for that distance.

Do brokers hunt stops?

A dealer that takes the other side of your trade has a conflict of interest. For NFA forex dealer members, Rule 2-43 forbids changing the price of an executed customer order except in narrow cases.

What is a liquidity sweep?

Another name for a stop run: price trades beyond a prior high or low, where resting stops add orders, then closes back inside the range.

Sources

  1. National Futures Association, Compliance Rule 2-43, Forex Orders
, Founder of PatternsWizard

Passionate about finance, quantitative trading and data analysis. Amaury built the PatternsWizard test bench, which detects each pattern on full price history and measures what happened next against random entries.

PatternsWizard tests chart and candlestick patterns on full market history, against random entries. How we test patterns.

179 patterns testedRandom-walk self-test passed

Related guides

All trading concepts articles