An order block is a price-defined zone used in Smart Money Concepts. One convention selects the last down candle before an upward move, or the last up candle before a decline. The rectangle is a testable chart hypothesis; it does not reveal participants, remaining orders or a required return to the zone.

What an order block actually is

Definitions vary by implementation: full High–Low range, candle body or a hybrid can bound the zone, while Wick, Close or midpoint conditions can retire it. LuxAlgo’s product calls a specified break “mitigation”; its concept page also uses that word for a return into the zone. Record first touch, response confirmation and invalidation separately rather than assume a universal definition or origin attribution.

“Price rose after this candle” is an observation; “a bank left unfilled purchases here” needs additional evidence. OHLC cannot reconstruct all participants’ books. A zone can be compared with support and resistance, but describing capital-flow intent does not make that intent observable.

Order blocks versus classic supply and demand zones

Supply/demand-zone and order-block rules select price areas differently. A narrower body or single-candle area changes the qualifying observations, not just the stop distance. Save the exact source candle and bounds before observing a later reaction.

Our independent EUR/USD model has t0 Low 1.0980,High 1.1004,Open 1.1000,Close 1.0990: full range 24 pips, body 10. Only the first touch can qualify, and its Close must exceed the zone top or observation ends. At t6, Low 1.1002/Close 1.1003/High 1.1010 touches the full zone but fails its Close condition; it never touches the body. At t7, Low 1.0995/Close 1.1006/High 1.1010 qualifies the body’s first touch, while the full zone already ended. A later bounce cannot revive the rejected full zone.

How traders mark and trade an order block

Here is our own long-only EUR/USD H1 research prototype, not a universal SMC standard or optimal parameter set. Use one provider’s Bid bars and fixed time zone. After down bar t0 require three consecutive up bars t1–t3, with t3 Close above the maximum High of the preceding 20 bars excluding t3. Only then save t0’s full range. Observe t4–t13 inclusive; any Bid below its bottom before entry invalidates it, and no touch by expiry ends observation.

Consider only the first bar whose range intersects the zone; its Close must finish strictly above the top, otherwise observation ends. After a qualifying close, the prototype buys at the next available Ask, sets SL 2 pips below the bottom, and a target at twice entry-to-SL distance. Skip if current Bid is at or below SL or the orders violate the provider’s conditions. One zone and one position at a time, with no adding or re-entry from that zone. The short side needs its own Ask-based execution description.

Specify any higher-timeframe trend and location filters before testing. A zone against the trend is not necessarily a losing trade, and nearby equal highs or lows do not prove remaining orders or improve success by themselves. The liquidity guide and Smart Money Concepts guide provide context, but the filters' value must be measured.

Spot FX is fragmented. A candle-defined zone is a hypothesis about past price structure, not a consolidated view of institutional orders.

A hypothetical example, step by step

Retain a separate, entirely hypothetical H4 illustration: a down candle opens 1.0830 with Low 1.0820; a later move reaches 1.0890. Choosing the Open–Low zone 1.0820–1.0830 gives 10 pips. This chosen convention becomes available only after its specified confirming move, not at the original candle. It is not evidence of bank inventory or a known favourable retest.

Assume a later actual Ask entry 1.0828, planned Bid stop 1.0815 and independent possible Bid exit 1.0905. Planned distance is 13 pips and target distance 77, not a guaranteed small loss or known liquidity pool. For 10,000 EUR and hypothetical USD 2 round-trip commission, exact-stop net is −USD 15 and target net USD 75; the gross ratio 77/13≈5.92 becomes net 75/15=5. A worse stop fill 1.0810 gives −USD 20. These are separate assumed outcomes, with financing/conversion omitted.

In the independent H1 prototype, t0 O 1.1000/H 1.1004/L 1.0980/C 1.0990 is not yet a confirmed zone. At t3, three up bars and Close 1.1045 versus prior 20 Highs≤1.1035 confirm 1.0980–1.1004. With no touch at t4–t5, t6 Low 1.0995/Close 1.1010 qualifies. Assumed next Ask 1.1012 and SL 1.0978 give 34 pips; target 1.1080 gives 68. For 10,000 EUR and commission USD 2, independent Bid exits 1.1080/1.0978/1.0973 produce net USD 66/−36/−41. Gross 2 becomes net 66/36≈1.83; spread is already included in fill sides.

A deleted zone still belongs to the sample

A disappearing rectangle should not remove its earlier event from the register. Compare the complete sample with a view retaining only rule-defined touches.

Scroll sideways to see the full diagram

Disappearing rectangles, permanent registerA and D meet the touch rule; B,C,F are invalidated and E expires. The full denominator is six zones; a display retaining A and D has denominator two.REGISTER STAYS, RECTANGLE GOESA — Rule-defined touchB — Invalidated before touchC — Invalidated before touchD — Rule-defined touchE — Observation window expiresF — Invalidated before touchFull register:2 of 6 =33.3%.Visible A and D only:2 of 2 =100%.Selection changes the denominator.

Fictitious model. Every one of six previously qualified zones enters the sample exactly once. Final classes are mutually exclusive: rule-defined touch, invalidation or expiry. No profit, win rate or broker offer is measured. TradingView — Strategies explains selection bias from omitting weaker cases; reviewing retained examples does not replace a complete sample.

Six hypothetical zones A–F were recorded before observation. The diagram does not show trade results.

This register's touch fraction is 2/6, not 2/2. A touch is not a successful trade: entry, execution and outcome require separate rules. If old zones are cleared from the drawing, retain their IDs, qualification times and final statuses in a table. Otherwise the current chart cannot reconstruct the sample available during observation.

The honest caveat: interpretation, not a view of orders

BIS’s December 2019 account of fragmented FX execution describes dealers temporarily warehousing client risk until opposing client flow offsets it. That does not reveal a dealer’s portfolio in a chosen candle. Individual venues can have order books, but no retail OHLC rectangle is a consolidated institutional-order view.

BIS Working Paper 405 studies information in customer currency trades and differences among customer groups in a historical dealer dataset. It does not identify those trades from your candles or validate this zone strategy. Distinguish observed price structure from claims about institutional intent.

Selecting a candle only after seeing the subsequent rally introduces hindsight and selection bias. Define the candidate, impulse, retest and invalidation rules at the time they become observable, then record every qualifying occurrence. A historical picture does not establish how often a live rule will fail or whether it has positive expectancy.

What to do tomorrow

Test the full zone lifecycle and its separate execution rule. Ten neat examples, thirty entries or a month on demo do not establish positive expectancy. Save all qualifying zones before their outcomes become known.

  1. Record every qualified zone. Save its ID, original bounds and confirmation time, then its touch, invalidation or expiry. Retain rejected and disappearing cases instead of selecting only surviving rectangles.
  2. Test a higher-timeframe filter. Classify the same candidates with and without the predefined trend rule, then compare net outcomes. Do not assume that discarding countertrend examples improves the remaining trades.
  3. Separate observation from fills. If one historical bar crosses both SL and target, OHLC alone cannot order the events. Use finer data or a declared conservative assumption; test complete rules after costs on a period not used to choose them. Demo fills do not guarantee live execution.