Source: https://forexiapro.com/research/how-often-does-a-liquidity-sweep-reverse/

# How Often Does a Liquidity Sweep Reverse? A Count on Real Days

Dylan Forexia11 November 20268 min read

How often does a liquidity sweep reverse? About one time in four, by a strict rule. Of 1,601 levels swept on seven dollar majors and gold from 6 April to 1 October 2026, 368 (23%) closed back, came back at least 40% of an average day and were not passed again by 17:00 New York. This counts price behaviour, not trades. Our own test of trading against the hunt lost money.

[![How often does a liquidity sweep reverse: EUR/USD on 27 May 2026, a sweep of the Asian high to 1.16614 followed by 39 pips down to 1.16221](/seo/how-often-does-a-liquidity-sweep-reverse/eurusd-liquidity-sweep-reversed-27-may-2026.webp)](/seo/how-often-does-a-liquidity-sweep-reverse/eurusd-liquidity-sweep-reversed-27-may-2026.webp)

EUR/USD, Wednesday 27 May 2026. The sweep of the Asian high is one of the 368 that reversed by our rule. We mark up a day only after it has reversed, so this chart flatters the idea.

A sweep is what we call [the Stop Hunt](/glossary/the-stop-hunt/) when the level is taken: price trades through a place where stops rest. The [guide to the liquidity sweep](/liquidity-sweep/) explains the idea. This page only counts what came next.

## How often does a liquidity sweep reverse, by each reading?

The answer moves with the word "reverse", which is why published figures disagree. Every level taken fell into one of four outcomes.

| What the level did after it was taken | Levels | Share |
| --- | --- | --- |
| No candle closed back by 17:00: it ran | 291 | 18% |
| Closed back, then a later candle traded past the wick | 825 | 52% |
| Closed back, came back 40% of an average day, wick not passed | 368 | 23% |
| Closed back, no such move, wick not passed | 117 | 7% |

So there are three honest answers, from loose to strict:

- **It closed back:** 1,310 of 1,601, or 82%. That is [the Close Back](/glossary/the-close-back/), and it is the number that makes sweeps look reliable.
- **It closed back and the wick survived the day:** 485 of 1,601, or 30%.
- **It also travelled:** 368 of 1,601, or 23%.

Of the 825 that were passed, 719 were passed before price had come back 40% of an average day. The other 106 came back that far and were still passed later the same day.

## What rule did we count, on what sample?

We used one rule for every market, and it is the same one behind the figures on our other pages. These are the terms.

- **Levels:** each edge of [the Asian Box](/glossary/the-asian-box/) and the previous trading day's high and low.
- **Taken:** the first 15-minute candle from the London open to 17:00 New York to trade past the level by at least 3% of the average daily range. A level counts once a day. The average is the 15 trading days before.
- **Close back:** a candle from that one on, the same candle included, that closes on the near side of the level or exactly on it, by 17:00.
- **The wick:** the furthest price reached by the close back.
- **Came back:** price moved 40% of the average daily range from the wick, read from the close back candle's close and then the far side of each later candle. We read it that way because the order of a high and a low inside one candle is not known. 40% is the line we use to call a day reversed when we mark it up.
- **Passed:** a later candle traded past the wick. A candle that both passed the wick and reached the 40% line counts as passed.

**Sample and dates.** Dukascopy's 15-minute bid candles on EUR/USD, GBP/USD, USD/JPY, USD/CHF, AUD/USD, USD/CAD, NZD/USD and gold, from Monday 6 April to Thursday 1 October 2026. That is 25 whole weeks and four days, 1,030 market-days, with $1 on gold counted as 1 pip.

**What is left out.** A previous day's level that Asia had already traded past. Everything after 17:00, so a turn the next morning does not count. Friday 2 October onward. The low of each day's first candle is clipped to its body, because it is often only the spread widening. Round numbers, equal highs and lows and trendlines have no count here.

## Does the answer change from market to market?

The strict figure ran from 18% on USD/CHF to 27% on EUR/USD. These are eight markets moving on one dollar in one stretch of months, so we would not rank them.

| Market | Levels taken | Ran | Closed back, wick passed | Came back 40%, wick not passed | Neither |
| --- | --- | --- | --- | --- | --- |
| EUR/USD | 218 | 36 | 107 | 59 | 16 |
| GBP/USD | 221 | 40 | 118 | 55 | 8 |
| USD/JPY | 183 | 34 | 85 | 43 | 21 |
| USD/CHF | 222 | 44 | 130 | 39 | 9 |
| AUD/USD | 180 | 40 | 83 | 41 | 16 |
| USD/CAD | 223 | 39 | 115 | 57 | 12 |
| NZD/USD | 170 | 25 | 91 | 37 | 17 |
| Gold | 184 | 33 | 96 | 37 | 18 |
| All eight | 1,601 | 291 | 825 | 368 | 117 |

The two kinds of level differ little. Box edges reversed by the strict rule on 286 of 1,179, previous-day levels on 82 of 422. A count of [which side of the Asian Box goes first](/research/london-open-fakeout-asian-range/) takes the box edges further.

## What did a reversal and a run look like?

One day shows a sweep that reversed and a sweep that was passed. EUR/USD on Wednesday 27 May 2026 had an Asian high of 1.16492. The 08:15 candle went through it, with a medium-impact ADP figure on the calendar in that candle, and the 08:30 candle reached 1.16614, 12 pips past. The 08:45 candle closed back at 1.16435. By the 09:30 candle price had come back 22 pips, 40% of that day's 55-pip average. Nothing passed 1.16614 again, and the low of the day was 1.16221 at 13:15, 39 pips under the wick.

We read that as buyers' orders and sellers' stops over the Asian high being used by whoever wanted to sell, and that is a reading. We cannot see who sold.

The other edge went differently later that same day. Without the first candle's wick our Asian low is 1.16279 (the chart's box keeps the wick, at 1.16250). The 10:45 candle went through it, the 11:00 candle reached 1.16223 and closed back at 1.16332, and the 13:15 candle traded to 1.16221, past that wick by a fifth of a pip. That one is in the 825. The rule is strict, and that is the point of a rule.

A run has no such moment. GBP/USD on Tuesday 21 July 2026 had an Asian low of 1.34203. UK jobs data had come at 02:00, and no high-impact US item was on the calendar. The 06:45 candle was the first past it by the 3% line. By 09:30 price was at 1.33596, 61 pips past, about three quarters of that day's 79-pip average. The 10:00 candle closed at 1.33987, its best close after that, still 22 pips short. The day ended at 1.33749. Anyone waiting for that level to reverse was still waiting at the close.

[![A liquidity sweep that did not reverse: GBP/USD on 21 July 2026, through the Asian low at 1.34203 to 1.33596, with no close back by 17:00 New York](/seo/how-often-does-a-liquidity-sweep-reverse/gbpusd-liquidity-sweep-no-close-back-21-july-2026.webp)](/seo/how-often-does-a-liquidity-sweep-reverse/gbpusd-liquidity-sweep-no-close-back-21-july-2026.webp)

GBP/USD, Tuesday 21 July 2026. Through the Asian low at 06:45, 61 pips past it by 09:30, and no close back by 17:00. It is one of the 291 that ran.

## Is the reversal count a reason to trade against the hunt?

No, and our own test says so. We ran the setup as a fixed rule that traded against each hunt, on 5-minute candles with real spreads and stated costs. Every year from 2021 to 2026 lost, between -0.10R and -0.23R a trade on about 2,200 trades a year. R is the amount risked on one trade. That is a different sample and a different rule from the count on this page, and the method and limits are in [our liquidity sweep backtest](/stop-hunt/liquidity-sweep-backtest/).

The candles on this page point the same way. A candle that was first past a level and closed back in that same candle moved a median of -1.0% of the average daily range the way it pointed after an hour, on 392 of them. That is nothing to lean on. A reversal seen afterwards is not a setup, and shapes such as [double tops and bottoms fail](/price-action/why-double-tops-and-bottoms-fail/) in the same way, by looking better afterwards than they did at the time.

## What does this count not prove?

It does not prove that a sweep causes a reversal, or that a sweep is a signal. The limits are these.

- **Candles, not trades.** There is no entry, stop, spread, commission, slippage or filter here. A share of levels is not a share of trades.
- **Past days only.** It covers one run of months in one setting of the clocks, on one bid feed. Another broker's candles will differ a little.
- **No yardstick.** Price also travels 40% of an average day with no sweep at all, and this count does not measure how often. So 23% cannot be read as better or worse than chance.
- **Not independent.** The eight markets move on one dollar, and a single candle can take levels on several of them.
- **Who did it cannot be seen.** Forexia reads these moves as market makers using known orders. The count shows where price went and nothing about who sent it there.
- **Other pages measure other things.** The 41% to 51% on [our stop hunt page](/stop-hunt/) is the share of breaks of the Asian range where the day's high or low printed past the range and price then reversed at least 40% of an average day, over 2021 to 2026. It is not the same count and should not be set beside this one.

More of these studies are on our [research page](/research/). Course 3 of ForexiaPRO, [Reverse Price Action](/pro), teaches this read step by step.

## What changed when this page came back?

This page was taken down on 31 October 2026 and put back on 11 November 2026. Its figures were rerun on 11 November 2026 with scripts/seo-counts.mjs. Every count matched the stored file exactly, to the candle. The sample still ends on 1 October 2026. No finding was reworded to read better than the numbers.

## Frequently asked questions

### Do liquidity sweeps always reverse?

No. Of 1,601 levels taken on seven dollar majors and gold from 6 April to 1 October 2026, 291 had no close back by 17:00 New York, and on 825 more a later candle traded past the wick after the close back. By our strict rule 368 reversed, about one in four. That is a count of candles, not of trades.

### How often does a liquidity sweep close back?

Most of the time: 1,310 of 1,601 levels closed back the same day, or 82%. A close back is not a reversal, though. After 825 of those 1,310, a later candle that day traded past the wick again.

### Is a sweep reversal rate a trading edge?

Nothing here shows one. This count has no entry, stop or cost and no yardstick for how far price moves without a sweep. Our separate test of trading against the hunt as a fixed rule lost money after costs in every year from 2021 to 2026.

## More from the blog

### [Copy Trading vs Forex Signals: What You Actually Get](/learn-forex/copy-trading-vs-forex-signals/)

Copy trading places the trades in your account; a forex signal leaves them to you. What each gives you and costs you, on one real dated EUR/USD signal.

### [Double Top Failure Rate: How Often Double Tops and Bottoms Fail, Counted](/price-action/double-top-failure-rate/)

The double top failure rate, counted: 983 tops and bottoms on 15-minute forex and gold candles, what the neckline did, and two real days that failed.

### [COT Report Release Date: When It Comes Out, With the 2026 Schedule](/cot-report/cot-report-release-schedule/)

The COT report release date is Friday 15:30 New York, positions counted the Tuesday before. The CFTC's 2026 dates, six holiday Mondays and the next four.