A liquidity sweep is not a breakout

A liquidity sweep is not a breakout. The difference is simple and it is mechanical. A sweep takes out a prior reference level by at least one tick and then closes back through that level. A breakout takes out the level and holds beyond it. One condition is a valid hourly manipulation in my model. The other is not. Confusing the two is how traders enter on the wrong side of a move, or force a trade on a day that never offered one.

I trade NQ and MNQ futures. The session boundary is 18:00 ET. Daily bias is built top down — daily, then 7-hour, then 4-hour — and locked at 10:00 ET. Entries only happen inside the 10:00–14:00 ET window. A valid setup needs an hourly manipulation first: price sweeps a prior reference low or high by at least one tick, then closes back through it. After that, a 5-minute confirmation triggers the entry. Stop goes one tick beyond the 5-minute manipulation extreme. Target is a static 2R. If the manipulation never happens, there is no trade that day. That is a normal outcome.

This post is only about the sweep. Not the bias. Not the 5-minute trigger. The hourly condition that has to print before anything else is allowed.

What a sweep actually is

A sweep is price running a level that other participants are defending or resting orders against, then failing to hold beyond it. In my rules the definition is tight.

Price must trade beyond a prior reference high or low by at least one tick. Then the same candle must close back through that reference. The wick beyond the level is the manipulation. The close back through the level is the rejection. Both parts are required. A wick alone is not enough. A close beyond the level is not a sweep — that is acceptance, and acceptance is breakout behaviour.

I use the hourly candle for this read. The manipulation has to print on the hourly. That keeps the condition slow enough to plan around and fast enough to matter inside the New York window.

Sweep versus breakout

A breakout and a sweep can look identical for most of the candle. Both send price through a level that mattered. The difference is where the candle finishes.

If price takes out a prior low, trades a few points beyond it, and then closes back above that low, the auction rejected the break. Stops below the low were likely triggered. The close back above says the market did not accept trade at those lower prices. That is a sweep of the low.

If price takes out the same low and closes below it, the auction accepted the lower prices. That is a breakout through the low, not a manipulation of it. I do not treat that as an entry condition for a long. The model is waiting for the failed break, not the successful one.

The same logic applies in reverse at a high. Sweep of a high: trade above the reference by at least one tick, close back below it. Breakout of a high: trade above and close above.

Traders get hurt when they treat every run through a level as continuation. Sometimes it is. Often, in the window I trade, it is the opposite — a grab of liquidity before the move the daily bias already pointed toward. The close tells you which one you got.

Why the close matters more than the wick

The wick is the probe. The close is the decision.

Anyone can point at a wick beyond a level and call it a sweep after the fact. That is hindsight. During the candle you do not know yet. Price can spike through a low, look like a collapse, and still close back above. Price can spike through a low and keep going. The close is the filter that separates those two outcomes inside my rules.

I do not anticipate the close. I wait for it. If the hourly is still open and price is beyond the reference, I have no manipulation yet. I have a candle in progress. Acting before the close is discretion. The model does not use discretion at the manipulation step.

This is also why the reference has to be defined before the candle starts. If you pick the level after you see the wick, you will fit the story to the chart. Write the level down first. Then let the candle either meet the condition or fail it.

Reference levels I actually use

A sweep needs something concrete to sweep. I do not use vague zones. I use levels I can mark with a horizontal line and a number.

Session highs and lows. Asia high and low. London high and low. These are clean because the session boundaries are fixed. Everyone watching those sessions can see the same prints.

Previous day high and low. The prior day's range is a standard reference. Stops and breakout orders cluster there. On the 4 August 2026 session the previous day on NQ had a high at 28964.25, a low at 28313.50, and a close at 28891.75. Those numbers are fixed before the new session opens. They do not move.

Prior swing points on the higher timeframes. The daily swing low that anchored the long bias on 4 August sat at 27201.50. That is a structural reference, not a session print. It informed bias. It was not the level I needed the hourly to sweep for entry.

Intraday candle highs and lows also matter once the New York morning is underway — specifically the hourly highs and lows that print before the entry window. Those become the near references the manipulation often runs.

I mark the relevant ones before 10:00 ET. I do not add new “references” mid-candle to rescue a setup that is not there.

The 4 August teaching moment

Tuesday 4 August 2026 is the cleanest recent example of why the level must be written down in advance, and why the instrument you read matters.

Bias resolved long off the daily swing low at 27201.50 and was locked at 10:02 ET. Asia had traded 28831.75 to 29098.75. London had traded 29007.50 to 29245.75. London never traded back into Asia’s range. The 25% retracement of London’s range sat near 29067. The London low at 29007.50 was the deeper reference. None of that is the sweep. That is context. The sweep question on the day was simpler: did an hourly candle take out a defined prior low and close back above it.

On MNQ the 08:00 candle low was 29186.50. The 09:00 candle traded to 29185.00. That is a sweep of 1.5 points beyond the 08:00 low. The same 09:00 candle closed at 29469. It took the low out by at least one tick and closed back through it. On MNQ, the hourly manipulation condition was met.

On NQ the same 09:00 candle low was 29191.25. The 08:00 low on NQ was 29188.00. The 09:00 candle on NQ did not take out the 08:00 low. It missed by 3.25 points. Same moment in the session. Two instruments. Opposite readings.

That is the lesson. MNQ gave a 1.5-point sweep. NQ gave a miss. If your reference was “the 08:00 low” without specifying the contract, you could talk yourself into either story. If you had MNQ in front of you, the sweep was real. If you had NQ in front of you, the sweep was not. The model does not allow you to switch charts after the fact to find the version that worked.

The session high later reached 29504.50. The day finished with no entry taken. Bias was long. Price moved up. The full entry condition still was not met under the rules I actually trade. No hourly manipulation that I accepted, no 5-minute confirmation stack I signed off on, no trade. That is process over feelings. A green day on the index is not the same thing as a valid setup in the model.

Why one tick is enough

The rule is at least one tick beyond the reference. Not a point. Not a “meaningful” break. One tick.

I do not grade sweeps by how far they travel beyond the level. A 1.5-point break through the MNQ 08:00 low counts. A deep plunge that still closes back through the level also counts. Distance beyond the level is not the filter. The close back through is the filter.

Traders add subjectivity here when they should not. They want the sweep to “look” like a stop run. Looks are not a rule. Either price traded beyond the level by at least one tick and closed back through it, or it did not. The MNQ 09:00 candle on 4 August is the example: 1.5 points beyond 29186.50, close back above at 29469. Qualifying sweep on that contract. No debate about depth.

What is not a sweep

A few patterns get mislabelled constantly.

A wick that touches the level but does not trade beyond it is not a sweep. You need at least one tick through.

A candle that trades beyond the level and closes beyond the level is acceptance. That is breakout or breakdown behaviour. It may be useful information for bias or for standing aside. It is not the manipulation condition I enter on.

A 5-minute sweep without the hourly condition is not enough. The model sequences hourly manipulation first, then 5-minute confirmation. Skipping the hourly step turns the method into something else.

A level you drew after the candle closed is not a reference. It is a narrative. References are marked before the test.

How this sits inside the rest of the model

The sweep does not replace bias. Bias comes first from the daily, 7-hour, and 4-hour work before the New York open. On 4 August that bias was long, locked at 10:02 ET, and not revised. The sweep is the hourly permission slip inside that bias. I am not sweeping lows to go short against a locked long bias. I am waiting for manipulation that fits the direction already defined.

After a valid hourly sweep, I still need the 5-minute confirmation before entry. Stop placement is defined from the 5-minute manipulation extreme, one tick beyond. Target is 2R from that risk. The day stops after +2R or after 2 losses. None of that activates if the hourly sweep never prints.

If the manipulation never happens, there is no trade. 4 August ended that way for me. The community sees this on the weekly New York livestreams often enough — sessions where the plan is clear, the bias is locked, and the tape never offers the condition. Repetition of the process matters more than forcing a position because the index moved.

Write the level down before the open

The MNQ versus NQ mismatch on 4 August only hurts you if the reference was fuzzy. If you wrote, before 10:00, “MNQ 08:00 low at 29186.50 is the reference I need swept and closed back through on the hourly,” you had a binary check. 09:00 low 29185.00, close 29469. Condition met on MNQ. If you wrote the NQ 08:00 low at 29188.00 as the reference, 09:00 low 29191.25 failed the test by 3.25 points. Condition not met on NQ.

Same clock time. Same market family. Different contract prints. The discipline is choosing the instrument and the level in advance, then accepting the print you get.

I trade the condition on the chart I am actually executing. I do not borrow a sweep from MNQ to justify an NQ entry, or the reverse. If you run both, define which contract’s levels govern the trade before the window opens.

What to do tomorrow

Before the New York open, mark the previous day high and low, the Asia high and low, and the London high and low on the contract you will trade. Add the prior hourly highs and lows that are relevant once the morning candles start printing. Write the numbers down. Lock your daily bias by 10:00 ET and leave it alone.

During 10:00–14:00 ET, watch for one thing on the hourly: a trade beyond one of those written levels by at least one tick, then a close back through the same level. If that candle closes still beyond the level, treat it as acceptance, not manipulation. If the hourly never sweeps and closes back through a reference, take no trade. Do not move the level. Do not switch contracts mid-candle to find a sweep that was not there on the chart you planned around.

Run that check tomorrow exactly once per hourly close inside the window. The result is either a valid manipulation to hand off to the 5-minute confirmation, or a blank session. Both results are the model working.