Study · EUR/USD · liquidity

When EURUSD Sweeps a High, Does It Reverse? Usually Not.

Instrument
EUR/USD
Theme
liquidity
Data window
EURUSD · 4,986 sweep events · 16 years of H1 data
Published
Reading time
6 min

Question & scope

The stop-hunt narrative says a wick beyond a swing high is a reversal signal. Across 4,986 EURUSD sweeps, only ~30% reversed — nearly 68% continued in the sweep direction. Most 'stop hunts' are just the level genuinely breaking.

Key finding · Historical sample

A EURUSD liquidity sweep — a wick past a swing level that closes back inside — reverses only about 30% of the time and continues ~68% of the time. Fading every sweep as a stop-hunt is a losing default. The reversal edge is a minority case you have to filter for, not assume.

Reversal rate by sweep depth (how far the wick pushed past the level)

Depth past the swing level · Reversal rate (%)

Reversal rate by sweep depth (how far the wick pushed past the level) — numeric data
Depth past the swing levelReversal rate (%)
0–3 pips29.9%
3–8 pips28.9%
8–20 pips30.7%
20–50 pips35%
>50 pips47.6%
A real EURUSD sweep reversal — 9 Jun 2022H1
sweep + rejectionswing high (swept) · 1.07482

Hollow: close ≥ open · Filled: close < openswing high (swept)sweep + rejection

Price wicked ~25 pips above the swing high with a large rejection wick, closed back inside, and reversed hard — the minority ~30% case where a sweep genuinely reverses instead of continuing. The strong wick and liquid NY hour are what tilt these odds.

Candle data
A real EURUSD sweep reversal — 9 Jun 2022 — candle data
TimeOpenHighLowClose
2022-06-08 17:001.074141.074311.072481.07282
2022-06-08 18:001.07281.072821.071631.07189
2022-06-08 19:001.07191.0721.071241.07153
2022-06-08 20:001.071551.071931.071281.07134
2022-06-08 21:001.071391.071681.071261.07159
2022-06-08 22:001.071631.071791.071321.0716
2022-06-08 23:001.071761.072041.071531.07186
2022-06-09 00:001.071861.072031.071561.07167
2022-06-09 01:001.071671.071781.070911.07119
2022-06-09 02:001.071191.071461.070941.07111
2022-06-09 03:001.071111.072341.071051.07224
2022-06-09 04:001.072241.073341.071991.07292
2022-06-09 05:001.072921.073171.072581.07264
2022-06-09 06:001.072611.072941.071851.07185
2022-06-09 07:001.071891.071971.070851.07143
2022-06-09 08:001.071411.071631.070171.07018
2022-06-09 09:001.070191.071741.069361.07152
2022-06-09 10:001.071521.07191.071091.0716
2022-06-09 11:001.071591.072411.071291.07186
2022-06-09 12:001.071861.074911.068771.0728
2022-06-09 13:001.072811.077361.072341.0734
2022-06-09 14:001.073421.074271.068071.06867
2022-06-09 15:001.068651.068691.064471.06488
2022-06-09 16:001.06491.06661.064531.0657
2022-06-09 17:001.065721.065971.064541.06522
2022-06-09 18:001.065221.065461.061791.06386
2022-06-09 19:001.063851.064361.062581.06261
2022-06-09 20:001.062651.062651.061221.0613
2022-06-09 21:001.061251.061991.061081.06174
2022-06-09 22:001.061521.06191.06151.06176
2022-06-09 23:001.061831.061971.061081.06152
2022-06-10 00:001.061531.062141.061261.06189
2022-06-10 01:001.061871.06241.061091.06237
2022-06-10 02:001.062371.062531.06171.06247
2022-06-10 03:001.062461.062981.062341.06294
2022-06-10 04:001.062951.063521.062821.06339
2022-06-10 05:001.063391.063491.062941.06305
2022-06-10 06:001.063061.064191.063041.06333
2022-06-10 07:001.063331.063461.061871.06297
2022-06-10 08:001.062941.064141.061971.06251
2022-06-10 09:001.062511.063291.06071.06143
2022-06-10 10:001.061431.061761.058981.05969
2022-06-10 11:001.05971.060621.058581.05929
2022-06-10 12:001.059331.05951.057091.05741
2022-06-10 13:001.057381.058581.052111.05221

Trading social media loves the "liquidity sweep": price spikes beyond a swing high, grabs the stops resting above it, then reverses. It's a compelling story. We tested it mechanically across every EURUSD occurrence in 16 years — a wick beyond a recent swing high or low that closes back inside — and measured what happened over the following day. 4,986 sweeps.

How we tested it

This public study is derived from OpenAlpha's internal TradingDesk EURUSD Sweep Reversal Atlas. A qualifying event is an H1 candle that wicks beyond an H4 swing high or low identified from the preceding 50 H4 candles, then closes back inside that level. We measure the outcome over the following 24 H1 candles:

  • Continuation: an H1 candle closes beyond the swept level.
  • Reversal: price moves one H4 ATR away from the level without an H1 close beyond it.
  • No clean resolution: neither condition occurs in the measurement window.

This is a historical classification study, not a forecast or a standalone trading signal.

The story is mostly wrong

OutcomeCountShare
Continuation (price closes beyond the level)3,40268.2%
Reversal (rejects and moves away from the level)1,50630.2%
No clean resolution781.6%

Continuation beats reversal better than two to one. Most of the time, a wick beyond a swing level is not a trap being sprung — it is the level genuinely breaking, with the close-back-inside being noise on the way through. Fading every sweep is a slow losing strategy.

Depth is the tell — but backwards from the folklore

If sweeps were stop-hunts, deeper grabs should reverse harder. They do the opposite of what most assume, and it's subtle:

Shallow pokes (a few pips past the level) reverse ~29% of the time — they're mostly continuation. Only the deepest sweeps (beyond ~50 pips) reverse close to half the time, and even then it's a coin flip. A deep sweep is more likely genuine exhaustion; a shallow one is usually just the level breaking. Either way, depth alone never gets you above 50% reversal.

Timing changes the odds more than depth

Where the sweep happens matters more than how deep it is:

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

Sweeps during active, liquid hours are far more likely to reject than sweeps in dead hours — where a poke past a level almost always just keeps going. Reversals need real two-sided participation to form.

What this means

We keep the exact reversal-entry filters private, but the transferable lessons are clear and they run against the popular narrative:

  • A sweep is continuation until proven otherwise. The base rate is ~68% continuation — start there, not at "reversal."
  • Depth cuts against folklore: shallow sweeps mostly continue; only deep, exhausted sweeps approach even odds of reversing.
  • Reversals are a liquid-hours phenomenon. In thin sessions, a level that gets swept is usually just gone.

The edge here isn't spotting sweeps — they're everywhere. It's refusing the reflex to fade them, and only taking the small, filtered subset where the odds actually flip.