How to Trade a Liquidity Sweep
To trade a liquidity sweep, mark an obvious level in advance and wait for price to exceed it. If price reverses back through that level within a stated number of bars, enter on the return with a stop beyond the extreme of the push.
A liquidity sweep is price pushing past an obvious level and reversing almost immediately. Mechanically it is a failed breakout. The added claim is about why it happened, and that part cannot be verified from a price chart.
Before you start
An obvious level marked in advance, because the sweep has to be of something. A prior high, a range edge, a round number — somewhere stops would plausibly sit.
A rule for how fast the reversal must come, or every failed breakout qualifies. Three bars, five bars, a number. Without it, the pattern is identified after price has already come back.
A stop beyond the sweep, which is usually wider than feels comfortable. That distance is the trade’s risk and it decides the size.
The steps
1. Mark levels where stops would plausibly sit
The high of a range, a prior swing, a round figure. The level has to be one many people can see, because that is the whole premise of the pattern.
2. Wait for price to exceed it
On this site’s shared series price traded through 85% of 39 twenty-bar levels, so this half of the pattern is entirely ordinary and carries no information on its own.
3. Require the reversal within your bar limit
Price back inside the level within your stated number of bars. Slower than that and it is a breakout that later failed, which is a different event with different odds.
4. Enter on the reversal, not on the extension
The setup is the return through the level. Entering while price is still beyond it is guessing that the sweep is a sweep rather than a breakout, and at that moment they are identical.
5. Put the stop beyond the sweep’s extreme
If price returns past that extreme, the reading was wrong. That is a structural level and it is often wide, which is information rather than an obstacle.
6. Size from that distance
Risk figure divided by the distance from entry to stop. A wide sweep produces a small position and the same amount at risk, which is the arithmetic that makes the correct stop affordable.
7. Log the ones that failed as well
Every level exceeded that never came back is a sweep that was not one. Recording those is the only way to know what proportion of your marked levels produce the pattern at all.
How to tell it worked
The level was marked before price reached it, in every case.
The reversal happened within your stated number of bars, measured rather than estimated.
The stop sat beyond the sweep’s extreme, so 1 adverse move costs your intended risk.
And levels that broke and kept going were recorded too, over at least 90 days.
The claim you cannot verify
That the move was deliberate, to trigger resting orders. Nothing in publicly available price data distinguishes a deliberate push from an ordinary breakout that ran out of buyers.
The weaker claim is the one worth trading. Price exceeded an obvious level and came back quickly, which means the move beyond it found no follow-through. That is observable, and it is enough.
Distinguishing it from a failed breakout
They are the same event. The difference is entirely in the speed of the return and in the story attached to the cause.
Which means the bar limit is doing all the work. A return within two bars is a different statistical event from one within twenty, and only the first is what people mean by a sweep.
And the story adds nothing you can act on. Whether stops were targeted or buyers simply ran out, the trade, the stop and the size are identical — so the narrative is a description rather than an input.
Which levels are worth marking
The ones a lot of people can see. A range high that has held twice, the previous session’s extreme, a round figure. The premise of the pattern is that orders cluster there, so an obscure level has nothing for price to sweep.
Levels that have been tested and held, rather than touched once. A boundary price has respected on two or three occasions is one people are watching; a single touch is a coincidence with a line drawn through it.
Not levels you derived from an indicator. A moving average or a calculated band is not somewhere resting orders accumulate in the way a visible high is, and the whole mechanism this pattern describes depends on that accumulation.
Four or five levels at a time is plenty. Marking twenty guarantees price is always near one, which converts a specific setup into a permanent condition and removes any ability to say no.
The original data
Of the 24,971 unique videos in research/search-study-corpus.jsonl, 69 mention liquidity sweeps in
the title, at a median of 17,537 views across 49 channels, and 64% of those titles are
instruction-shaped. Order blocks appear in 391 at 2,786 and fair value gaps in 150 at 28,170. The
counts come from site/corpus_count.py.
69 videos at 17,537 across 49 channels. A moderately covered concept with a large audience per video, sitting on top of a pattern that has been described as a false breakout for decades under a name with no claim about intent in it.
The answer to the question on that chart depends on the number you wrote down. If your limit was three bars, this is a failed breakout rather than a sweep — and stretching the limit to include it is how a rule with a testable boundary becomes a label applied to whatever happened.
When it fails
The failure is the sweep identified after the return, and it makes the pattern look infallible. Every level that price exceeded and came back from is a sweep in review. The levels price exceeded and kept going past are not recorded, because nobody labels those. The pattern therefore has a perfect historical record, assembled entirely by discarding the cases where it did not apply — and on this site’s series price traded through 85% of 39 twenty-bar levels, which is the size of the discarded set.
The second failure is no bar limit. Any failed breakout then qualifies.
A third is entering during the push. At that moment the two events are identical.
A fourth is a stop inside the sweep. The extreme is the invalidation.
A fifth is trading unmarked levels. The level has to be obvious in advance.
And a sixth is trading the narrative. Intent is not visible on a chart.
Related
Liquidity sweep covers the pattern in full. Stop hunts is the same claim stated more directly. And false breakout is the same event described without the intent.
The honest framing is that this is a failed breakout, which is a pattern that existed long before the vocabulary around it. Whether stops were deliberately targeted is unknowable from a chart. What is knowable is that price exceeded a level and came back, and that is enough to trade without the narrative.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.