Forex Liquidity Sweeps: Wick Reclaims and Failed Breakouts

1. Separate the observation from the story

Traders use liquidity sweep, stop run and false breakout with different meanings. Some refer to a wick beyond a swing; others require a close back inside, an opposite structure break or a later entry pattern. State your conditions before comparing examples. A familiar name is not a complete definition.

Here, a wick reclaim means that a completed candle breaches one frozen boundary and closes strictly inside the reference pair on that same candle. A closed break means a strict close outside. A later return means that a subsequent completed candle closes strictly inside the original pair within a preset window. These are price observations, not direct measurements of liquidity.

The practical question is what was known at each close. Calling a move a failed breakout after seeing a later return must not change what you could have recorded at the original break. This distinction matters when reviewing a chart, annotating an EA signal or comparing a backtest with live behaviour.

2. What liquidity means, and what a candle cannot reveal

In market microstructure, liquidity concerns the ability to transact, including available prices, depth and the cost of execution. The word does not mean every visible high contains a known quantity of stops. A retail OHLC candle records extrema and endpoints for its feed; it does not enumerate resting orders across the market.

The BIS describes spot FX as an opaque, fragmented OTC market. Our inference is that a single broker chart is an incomplete view of market-wide order placement. A dealer or venue may have additional information, but that information is not supplied by the candle in this exercise.

A sweep-shaped candle can support the statement “price crossed this reference and finished inside.” It cannot, by itself, identify who traded, how many stops were triggered, whether an institution deliberately targeted them, or whether the next move will reverse. Keep those claims separate from the recorded price event.

3. Why stops are relevant without proving a stop hunt

Carol Osler’s 2002 New York Fed study reports evidence that clustered currency stop-loss orders can contribute to price cascades. This supports a possible order-flow mechanism; it does not establish the cause of a particular retail-chart wick or validate the rule used below.

A boundary can be useful as a reference even when the actual order distribution is unknown. Equal-looking highs, round numbers and previous session extremes are candidate landmarks, not a verified map of other traders’ orders. Changing the landmark after seeing the excursion destroys the comparison.

Prefer a neutral journal entry such as “upper reference breached by the high; close returned inside” over a causal claim such as “banks collected stops.” The first can be checked against prices and timestamps. The second requires evidence about orders and participants that this guide does not possess.

4. Freeze the reference before the candle opens

Use the earlier Market Structure guides’ confirmed pair: R_H = 1.1040 and R_L = 1.0960, known at 11:00 on 2026-10-09. The classified snapshot is HH/HL. Those guides explain how strict five-bar pivots become known only after their right-hand candles close.

For the shared 11:00–12:00 candle, the invented Bid prices are O = 1.0974, H = 1.1032, L = 1.0972, C = 1.1030. At 12:00, the close is inside the frozen pair. Both following alternatives therefore begin with the same available information.

Keep the symbol, feed, H1 timeframe, broker server clock and pivot identities unchanged. References remain frozen throughout this exercise, including the follow-up window. A newly confirmed pivot may belong to a later exercise; it cannot replace the level against which an earlier event is being judged.

5. Define the strict tests and the observation window

All prices below are Bid. Strictly inside means R_L < C < R_H. A close exactly on either boundary is a boundary close, not an inside reclaim. A touch by an extreme is not a strict breach. For this example, one pip is 0.0001.

While waiting for the first closed break, require the previous close and the new candle’s open to be inside or on the frozen pair. If the open gaps outside, record a gap and stop this exercise instead of presenting it as an ordinary inside-to-outside test. If both extremes breach both boundaries on one candle, mark the bar ambiguous: OHLC does not reveal which side was visited first.

An upper wick reclaim requires H > R_H and R_L < C < R_H, with no lower-boundary breach. A lower wick reclaim requires L < R_L and the same strict inside close, with no upper-boundary breach. A wick observation leaves the pair available for a later closed-break test.

The first eligible C > R_H or C < R_L starts a closed-break episode. Inspect only the next N = 2 consecutive H1 candles. The first strict inside close is a return within the window. A close beyond the opposite boundary ends the episode as a separate opposite break. Otherwise, a follow-up candle breaching both boundaries is ambiguous. Either ends this exercise without a qualifying inside-return label.

• Count the window from the candle after the break, not from the break candle itself.
• A boundary close uses one window position but does not qualify as a return.
• After two valid follow-up candles without a qualifying return, record NO_RETURN_IN_2 and end the episode.
• If a required candle is missing, record INCOMPLETE_WINDOW rather than substituting the next two available candles.

6. Path A: a same-bar wick reclaim

At the 13:00 close, the high 1.1046 is above 1.1040, but the close 1.1037 is strictly inside the frozen pair. The excursion is (1.1046 - 1.1040) / 0.0001 = 6 pips; the close is 3 pips below the upper reference. The lower boundary has not been breached.

Record an upper wick reclaim at 13:00. There is no closed breakout to fail on that candle, and no BOS under the preceding guide’s strict-close convention. The later inside closes add follow-up context without turning the original wick into a confirmed reversal.

These are synthetic bars with continuous opens. The scheduled closing time names when a completed candle can first satisfy the observation; live recognition also depends on the arrival of the next quote.

7. Path B: a closed break, then a timed return

The alternative 13:00 close is 1.1042, or 2 pips above the same upper reference. Record the closed upper break at that time. With the frozen HH/HL context, it is also a continuation BOS under the preceding guide’s convention.

The 14:00 close 1.1041 remains outside, so the first follow-up candle does not qualify. At 15:00, the close 1.1036 is strictly inside, on the second following candle. Record a separate return within the preset window, 4 pips below the upper reference.

This episode meets our descriptive failed-breakout condition: a first closed break followed by an inside close within the fixed window. The later return does not erase the earlier BOS entry. At the original break, the return was still unknown. No trade, entry price or profit is implied.

8. Read the timeline in the diagrams

The panels share the same references and the same starting close at 12:00. Only the next three candles differ. The small white markers indicate the closes; the wick tops alone are insufficient to distinguish the two first observations. Read the close labels and scheduled times together.

The lower reference is outside the plotted scale in both panels. The chart zoom makes the upper-boundary comparison legible; it does not depict the full distance between the pair. No shaded order pool or volume estimate is drawn because neither is part of the available data.

9. Boundary cases that change the label

Use a deterministic label for awkward candles instead of deciding by appearance. The following cases assume an eligible start and the same frozen references; follow-up cases use the upper-break episode.

For a lower break, mirror the comparisons around the lower boundary and keep the inside test unchanged. A return above the lower reference is not sufficient if the same close is above the upper reference; that is an opposite break, not an inside return.

10. Keep Bid observations separate from Ask conditions

Standard MT4 historical charts use Bid; buying and short-position stop conditions involve Ask. A Bid-only candle cannot establish every order trigger. Consult the order type and execution record rather than assuming that a chart wick reproduces an Ask path.

For an illustrative single quote, Bid = 1.1039 and Ask = 1.1042 give 3 pips spread. Ask is above 1.1040 while Bid is below it. An Ask-side threshold may therefore be reached without a Bid breach at that quote. This is a separate quote example, not the reconstruction of either H1 path.

For another illustrative quote, Bid = 1.1042 and Ask = 1.1044 give 2 pips spread. Neither example supplies a guaranteed fill. With five decimal places, Point = 0.00001 and the 6 pips excursion equals 60 points. Preserve the pip definition instead of transferring it unchanged to other symbols.

11. Reconcile the event with BOS and CHoCH

The first guide describes confirmed highs and lows. The second describes a closed break relative to a frozen structure snapshot. This guide describes the relation between an excursion, its close and subsequent closes. These observations answer different questions and can coexist in the same journal.

An upper wick reclaim in an advancing snapshot leaves the strict-close BOS condition unmet. A closed upper break satisfies that condition; a later inside return records what happened next. Neither automatically supplies a new LH/LL classification or an opposite CHoCH.

If the close instead crosses the frozen lower boundary, record the opposite closed break at its own timestamp. Reassess structure using the declared pivot method in a new exercise. Do not relabel the full price path retrospectively just because the final screenshot looks like a reversal.

12. Recognize the close without borrowing future data

In an MT4 timeseries, the forming bar is shift 0; after a new bar appears, the preceding completed bar is shift 1. Evaluate the completed candle once per symbol, timeframe and opening timestamp. Repeated ticks must not duplicate the same closed-break or return record.

Store both the scheduled close and the actual observation time. A clock reaching 13:00 does not prove that a new quote arrived then. A later observation cannot be used as evidence of an order filled at the earlier historical close.

Check the required H1 timestamps and data completeness before counting the window. Use integer price units or a declared tick-size policy for strict comparisons in code; binary floating-point noise must not turn an equality into a breach. Freeze that policy before testing the observations.

13. Build a journal that another reader can reproduce

Keep the original event and the follow-up in separate fields. The following record is sufficient to audit these synthetic exercises without knowing the next candle in advance.

1. Record the symbol, feed, price side, timeframe, clock, reference values, pivot identities and the time each reference became known.
2. Save the pre-bar previous close and open; flag an outside opening or a missing candle before applying the event tests.
3. At each completed candle, record its OHLC and label a touch, boundary close, eligible wick reclaim, closed break or ambiguity using the fixed rules.
4. At the first closed break, save its direction and timestamp and start N = 2. Preserve any earlier wick observation as a separate event.
5. Record each required follow-up close and its window position. End with the first inside return, opposite break, ambiguity, NO_RETURN_IN_2 or INCOMPLETE_WINDOW.
6. Keep the event record separate from any proposed trade. If testing execution, add the observed quote, entry rule, stop, exit, costs and size before reviewing outcomes.

14. Measure the observation before inventing an edge

A screenshot selected because price later reversed is a biased sample. Count every eligible episode under unchanged references, boundaries and window length, including no-return and incomplete cases. Do not keep extending the window until the return appears.

A useful first statistic is the fraction of complete, non-ambiguous closed-break episodes that return strictly inside within the fixed window. State the denominator and report opposite breaks and incomplete or ambiguous observations separately. That statistic describes the label; it is not a trading win rate.

A later trading study requires predefined entry and exit rules, spreads, slippage, sizing and untouched evaluation data. These examples test event recognition only. Forex trading can lose capital; a sweep or failed-breakout label does not establish a profitable strategy.

15. Questions that often cause confusion

Does a liquidity sweep prove that stops were triggered?

No. Our candle rule identifies a price excursion and close. Identifying actual stop orders requires additional order or execution evidence.

Is a wick reclaim the same as a failed breakout?

Under this guide’s convention, they are separate. A wick reclaim finishes inside on the breach candle; a failed-breakout episode first has an outside close and then an inside return within the preset window.

What if the return arrives on the third following candle?

It falls outside N = 2. Keep NO_RETURN_IN_2 and record the later return separately. Changing the window afterward introduces hindsight.

Does returning inside cancel the original BOS?

No. The original closed-break event remains dated when it occurred; the inside return is a later observation. A new structure classification requires its own evidence.

Can the same rules be used below a low?

Yes, mirror the strict breach and break directions. Keep the same inside test, timing, ambiguity checks and fixed follow-up window. This does not imply that either direction has a trading advantage.

16. Sources and scope of the example

The primary sources below support the market-structure and platform distinctions stated in this guide. The frozen-level rules, two-candle window and synthetic paths are our explicit teaching convention; these sources do not validate their profitability.

• BIS Working Paper 1094 — The foreign exchange market
• New York Fed Staff Report 150 — Stop-Loss Orders and Price Cascades in Currency Markets
• MetaTrader 4 Help — Order Types and Bid/Ask distinctions
• MQL4 Reference — Indexing Direction in Arrays, Buffers and Timeseries