Liquidity Sweep Trading: How Stop Hunts Set Up Reversals

A liquidity sweep is when price pushes beyond a recent swing high or low — far enough to trigger the stop-losses and breakout orders resting there — then closes back inside the prior range. The popular narrative is that this is always a trap: "smart money" grabs the stops, then reverses the market in the opposite direction. It's a compelling story, and it's mostly wrong. We tested it directly on EUR/USD, and the real numbers change how a sweep should be traded.

What a liquidity sweep actually is

Define it mechanically, not narratively: a wick pushes past a recent swing level, then the candle (or a following candle) closes back on the other side of that level. That's the entire pattern — no assumption about intent required.

The reasoning behind why it happens is straightforward market structure, not conspiracy: stop-losses and breakout entries cluster just beyond obvious swing points, because that's where traders place them. A cluster of resting orders is liquidity — it's easy for a large order to execute against without moving price much further. Price is drawn toward areas of resting liquidity because that's where size can transact.

Where the popular version goes wrong is the next step: assuming that taking the liquidity always means the move is done and about to reverse. Our own data says otherwise.

What actually happens: 4,986 sweeps, tested

We identified every occurrence in 16 years of EUR/USD hourly data where a wick pushed past a recent swing high or low and closed back inside, then tracked what happened over the following day.

Outcome	Share
Continuation (price closes beyond the level)	68.2%
Reversal (rejects and moves away from the level)	30.2%
No clean resolution	1.6%

Continuation beats reversal by more than two to one. Most sweeps are not a trap — they're the level genuinely breaking, with the close-back-inside being noise on the way through rather than a rejection. Fading every sweep as an automatic stop-hunt reversal is a losing default, not an edge.

Depth is backwards from the folklore

If sweeps were mechanically stop-hunts, you'd expect deeper pushes past the level to reverse harder — a bigger grab, a bigger snap-back. The data shows almost the opposite:

Depth past the level	Reversal rate
0–3 pips	29.9%
3–8 pips	28.9%
8–20 pips	30.7%
20–50 pips	35.0%
Beyond 50 pips	47.6%

Shallow pokes — a few pips past the level — reverse only about 29% of the time; they're mostly continuation. Reversal odds only climb meaningfully once the sweep is deep (beyond ~50 pips), where a genuinely exhausted move is more plausible. Even then, it's close to a coin flip, not a high-probability setup. A shallow wick past a level is weak evidence of anything.

Timing matters more than depth

The clearest signal in the data isn't depth — it's when the sweep happens:

Session window	Reversal rate
London / New York hours (07:00–16:00 UTC)	~35%
Late / thin hours (20:00–21:00 UTC)	14–18%

Sweeps during liquid, two-sided hours are meaningfully more likely to actually reject. In thin, low-participation hours, a level that gets swept is disproportionately just gone — there isn't enough opposing volume left in the session to force a rejection. A 3am sweep and a 10am London sweep are not the same signal, even if they look identical on the chart.

How to actually trade the concept

Given the base rate is continuation, treat a fresh sweep as continuation until proven otherwise, and require real evidence before assuming reversal:

1. Default to continuation. The unconditional base rate is 68% continuation. Don't start from "reversal" and look for confirmation — start from "continuation" and look for a reason to override it.
2. Weight session timing heavily. A sweep during London or New York hours carries meaningfully more reversal weight than the same pattern in the Asian session or late-day thin hours.
3. Don't treat depth alone as a signal. Only genuinely deep sweeps (well beyond typical noise for the instrument) approach a coin-flip reversal rate — shallow pokes are closer to noise than to a stop hunt.
4. Require confluence, not just the sweep. A sweep that lines up with another independent signal — a fair value gap, a round number, a session boundary — is a stronger case than a sweep alone. This is the same confluence principle our level-cluster research found: an isolated signal resolves close to a coin flip, while a genuine cluster of independent signals meaningfully outperforms.
5. Size for the base rate, not the narrative. If you're trading reversals off sweeps, you're trading the minority case (~30%) by design — position size and stop placement should reflect that you need a real edge-filter to get above the base rate, not the sweep alone.

What this means for your prep

Liquidity sweep trading isn't wrong as a concept — it's incomplete as usually taught. The stop-hunt narrative correctly identifies where the market is likely to go (toward resting liquidity beyond a swing point). It gets what happens next backwards by defaulting to reversal when the data defaults to continuation. Build your prep around the base rate, then use timing and confluence as the filters that occasionally flip the odds in reversal's favor.

See it live, and go deeper

The full study — when EURUSD sweeps a high, does it reverse? — has the complete breakdown across all 4,986 events, including how depth and timing interact. It's the direct evidence behind everything in this guide.

To build this into a repeatable session process, start with how to prepare for a EURUSD trading session, and read what is a fair value gap next — sweeps and fair value gaps are usually discussed together in ICT methodology, and the confluence caveat in this guide applies to both.

If you'd rather see how sweeps get read in a live session prep than build the filter yourself, join the free journal — every EURUSD, gold, and S&P 500 prep is published with the reasoning shown before the session starts.