Liquidity sweep

The observable event is simple and contains no mystery: price traded above a prior high, or below a prior low, and then traded back through it. That is the entire chart-level fact. Everything else attached to the term is interpretation, and separating the two is most of what this page is for.

The interpretation goes like this. Stop-loss orders cluster just beyond obvious levels, because that is where a great many traders put them. Resting orders are what somebody wanting size needs in order to fill. So a poke beyond the level fills those orders, and price returns once they are done. It is a coherent story, and it may even be true. It is not something a candle chart can confirm.

  equal highs at 2341.5 and 2341.7  (an obvious level)

  ... price trades to 2343.2         <- beyond
  ... closes back at 2338.4          <- and returns

  observed:    traded beyond, came back
  NOT observed: whose stops, how many, filled by whom

Why "stop hunt" is the wrong word for what you can see

"Hunt" names a motive. A retail chart carries executions: prices at which trades happened, and volume only for centralised venues. It does not carry resting orders, participant identity, or intent. So the sentence "they ran the stops" describes something nobody watching a chart is in a position to observe, and repeating it turns a guess into an assumption that then never gets tested.

There is also a plainer explanation available for the same candles. Obvious levels attract orders of every kind — breakout entries, stop losses, take profits, options-related hedging. Price moving quickly through a zone with a lot of activity and then reverting is what any market does around a level that many participants care about, with no coordination required.

Defining a sweep well enough to count

• Which level. A single swing high, two roughly equal highs, a session high, a prior day's high — these are different populations with different frequencies.
• How equal is equal. Equal highs are almost never exactly equal. You need a tolerance, in points or ATR, and it changes how many events you find.
• How far beyond. Any tick past, or a minimum distance? A one-tick overshoot and a thirty-point overshoot are being counted as the same event otherwise.
• How fast back. Within the same candle, within n candles, or no limit at all. Without a limit, every level that is ever revisited eventually qualifies.
• Wick or close. Whether a close beyond the level disqualifies it from being a sweep.

The fourth one is where most informal testing quietly fails. "Price swept the high and reversed" is trivially true of any level given enough time, so a sweep with no time limit is not a pattern — it is a description of a chart that eventually moved.

The part that is genuinely yours to measure

Whatever is happening in the market, something specific and measurable is happening to you: sweeps are one of the commonest ways a stop is taken before an idea works. That is not a claim about intent. It is a fact about placement, and it is visible in your own record.

Sage reconstructs how far each trade went against you before it went anywhere useful — see MAE and MFE. If your losing trades cluster at a maximum adverse excursion just past an obvious level, the conversation to have is about where your stop sits relative to those levels, not about who moved price. The position size calculator turns a different stop distance into the position that keeps your risk the same.

And the part that is about you rather than the chart

Being stopped just before a move works is one of the most reliable triggers for the behaviour that costs traders far more than the original loss did. A stop taken at an obvious level feels personal in a way an ordinary loss does not, and the trade that follows is frequently larger, faster and unplanned.

This is worth tagging rather than remembering. Sage's psychology analytics compare what you did after trades you labelled this way against the rest of your record, with the sample size attached — and the guide to tilt and revenge trading covers the pattern in more detail.

Where traders disagree

This vocabulary is not standardised. These are the live disagreements — worth knowing about before comparing your results with anyone else’s.

• ?Whether sweeps reflect deliberate targeting of stop orders or ordinary activity around levels many participants are watching — two readings that produce identical candles.
• ?What tolerance makes two highs "equal", and whether the level must be equal highs at all rather than any prior swing.
• ?How quickly price must return for a move beyond a level to count as a sweep rather than a breakout.
• ?Whether a close beyond the level disqualifies the event, or whether only the subsequent return matters.

Test it on your own trades

Whether this holds is a question about your record, not about the term. These are the steps that make the answer trustworthy.

1. 1Fix the level type, the equality tolerance, the minimum overshoot, the return window and the wick-or-close rule in writing before counting anything.
2. 2Count every level that met the definition, including the ones price broke and kept going — those are the denominator.
3. 3Read your own maximum adverse excursion on losing trades and check whether it clusters just beyond obvious levels; that is a stop-placement finding you can act on.
4. 4Compare expectancy in R on sweep-tagged trades against your baseline, with both sample sizes shown.
5. 5Tag the trade that follows a stop taken at an obvious level, and look at what your record says about those separately.

You will also need

• Market structure shiftA reading of price as a sequence of swing highs and lows, where a break beyond the last swing either continues the sequence (BOS) or reverses it (CHoCH).
• Order blockThe last candle against the direction of a sharp move, marked as a price zone and watched for when price returns to it.
• Premium and discountA range divided at its 50% level — price above the midpoint is premium, below it is discount, and the midpoint itself is equilibrium.