Trading for a gap to fill assumes a return to a previously defined level. A blank part of a chart does not establish whether or when that return occurs, or whether a position profits after costs. Define the gap, fill horizon, price basis and order sequence before counting outcomes. The fictional examples below separate chart measurements from executed trades.

What a price gap actually is on a chart

An opening gap can be measured as new Open minus previous Close in one series. A full-range down gap instead requires new High below previous Low; an up gap requires new Low above previous High. These are different tests. A jump in one provider’s quotes or missing records does not establish that no venue traded between those prices.

Check the instrument and provider’s session calendar, price component and data completeness. OANDA US publishes daily breaks and instrument exceptions as well as weekend hours; it is one provider schedule, not uninterrupted worldwide Forex. OANDA’s candle API also distinguishes raw opening price from smoothing with the previous close. Smoothing changes an opening-gap calculation. Compare Bid with Bid or Ask with Ask, and keep midpoint chart touches distinct from available buy or sell quotes.

Four classical gap types

Technical analysis commonly names common, breakaway, runaway and exhaustion gaps. They describe selected context and interpretation, not automatic orders or a future deadline. Stock-chart examples do not supply a verified occurrence or fill rate for a currency strategy. An “exhaustion” label assigned after the reversal cannot be assumed known at the opening.

Conventional context labels; no guaranteed fill or entry
Common gapA gap within a trading range under a chosen context rule
Breakaway gapA gap outside a previously defined range
Runaway gapA gap within an existing directional move; “halfway” is an interpretation
Exhaustion gapA gap interpreted as near a move’s end; later reversal may be part of its recognition

Measure the gap and the chosen fill level

Our hypothetical EUR/USD Bid candles use one pip = 0.0001. Previous O/H/L/C is 1.0990/1.1030/1.0980/1.1000. Scenario A’s new candle is 1.0970/1.0990/1.0950/1.0985: its opening gap is −30 pips, but full ranges overlap from 1.0980 to 1.0990. Independent scenario B is 1.0960/1.0970/1.0945/1.0965: opening gap −40 pips, while the full-range gap is only 10 pips from 1.0970 to 1.0980. Neither new High reaches the old Close 1.1000.

Returning to 1.0980 would fill B’s full-range gap but not its opening-to-old-Close difference. State which event you count. The complete new High/Low are known only after the bar, so they cannot filter an order supposedly placed at its open. These are two alternative manufactured bars, not market research or a liquidity “magnet”.

A stop, size and cancellation rule belong to the tested plan. Neither a small displayed gap nor a 1% input establishes an acceptable universal allocation or maximum realised loss. Ordinary stop activation and execution are distinct; the fill need not be the first quote you saw. A guaranteed-stop feature has its own provider, eligibility and cost terms. Where applicable retail-CFD rules provide account-level negative-balance protection, covered liability is limited to dedicated account funds; this does not preserve those funds or guarantee a stop fill.

“Filling the gap” and percentage change

Returning to the same price does not mean reversing the move by the same percentage. The denominator changes.

Scroll sideways to see the full diagram

“Filling the gap” and percentage changeOur model: prior Close 100, new Open 90 in arbitrary units. A 10-unit fall from 100 is 10%. A hypothetical return from 90 to 100 needs a 10-unit rise from 90, or about 11.11%. Both levels share a linear price scale. The recovery arrow is an assumed scenario, not an observed future market.Equal 10 units, different percentagesOur model: Close 100 → new Open 90.10090−10%+11.11%Base: 100Base: 90Downward gapAssumed returnA scenario, not a forecast.

Assumptions: Our fictional levels of one positive price, without spread, costs, leverage or trades. The gap compares previous Close = 100 with next Open = 90. “Return to 100” is only a controlled arithmetic scenario, with no assigned deadline or probability.

Gap: (90 − 100) / 100 × 100% = −10%. Return: (100 − 90) / 90 × 100% ≈ +11.1111%. By contrast, 90 × 1.10 = 99. Equal 10-unit distances do not mean equal percentage changes: one base is 100, the other 90.

OANDA OHLC documentation distinguishes the period’s first and last prices. Our calculation does not prove a gap will fill or describe an actual position’s result.

After 100 → 90, a 10% rise reaches only 99.

How to trade the weekend gap (and when to avoid it)

In a separate fictional fill model, previous Bid is 1.1000 and the new quote is Bid 1.0970/Ask 1.0974. Assume a full buy of EUR 10,000 at 1.0974 on a USD account and USD 2 total commission. One pip is USD 1. Returning Bid to 1.1000 gives 26 pips from the Ask entry, or USD 24 after commission, rather than a 30-pip net gain.

Alternative full-sale outcomes in the fictional fill model
Actual assumed sale BidGrossAfter commission
1.1000, old-Close target+USD 26+USD 24
1.0955, exact assumed stop fill−USD 19−USD 21
1.0945, worse stop fill−USD 29−USD 31

The last scenario assumes a stop trigger 1.0955, Bid jumping from 1.0957 to 1.0947 and an actual fill only at 1.0945. Trigger-to-fill difference is 10 pips. This is a specified sequence, not a promise of the first available quote for every order. Spread is already in the Ask/Bid fills; financing, conversion and other fees are omitted. If the stopped-out price later returns to 1.1000, that earlier position remains closed at a loss.

For a hypothetical gap-fill test, define the feed, reopening time, entry delay, target near the previous close and invalidation stop before observing the outcome. If the target is not reached by a stated time, close at the available market price; a time exit cannot promise a return to the entry price. Check the provider's terms for any guaranteed-stop product and its eligibility and cost.

Record scheduled events as context or a predefined test filter. An empty calendar does not exclude an unexpected announcement, and skipping a listed election does not guarantee capital preservation. Provider terms matter too: IG UK’s guaranteed-stop feature fixes the exit level under its own terms, charges a premium when triggered and has editing/minimum-distance restrictions. This does not establish availability for all brokers, products or accounts.

Breakaway, runaway and exhaustion gaps in practice

A retrospective classification is not a real-time signal. To compare gap-fill and continuation hypotheses, freeze the context definition, availability time, entry, protective exit and horizon separately. A subsequent opposite gap can motivate an island-reversal interpretation, but no “strongest signal” ranking or automatic order follows. Size or activity filters need identified data and later evaluation; OTC tick counts are not an entire exchange’s transaction volume.

Three classic mistakes in gap trading

Keep three test errors separate: future information in the gap label, incorrect quote-side execution and incomplete observation of the fill horizon. Our next example demonstrates the third without estimating market probabilities.

  • Event context: define a calendar filter beforehand, retain excluded cases and avoid treating absence of listed news as a guarantee.
  • Definitions: do not switch opening-gap and full-range targets or assign a future-confirmed exhaustion label to an earlier signal.
  • Execution: distinguish trigger from fill, calculate quote-side costs and verify any guaranteed-stop product’s specific terms.

What to do this weekend

Our six fictional gaps use a predefined target-touch horizon of ten completed bars. A touches at bar 2, B at 6, C has ten observed bars without a touch, D has only three observed bars without one, E only one, and F first touches at 12. A/B qualify; C/F do not meet the ten-bar horizon; D/E are unknown. Among four resolved cases 2/4=50%, but that is not automatically the full six-case rate. With the two unresolved outcomes, the six-case proportion lies between 2/6 and 4/6, about 33.3–66.7%.

NIST describes right censoring when later event times remain unknown after observation ends. We transfer that principle to this fictional fill-time exercise without fitting a market model. Mark incomplete follow-up separately rather than automatically count it as failure. A final price touch also differs from profit: in another fictional sequence five outcomes +USD 24 and four −USD 31 total −USD 4 despite five of nine profitable trades.

  1. Choose the feed, quote side, session calendar, gap definition, target and horizon. Retain incomplete observations rather than silently count them as failed fills.
  2. Define any event exclusions before outcomes and record them. Scheduled-news filters do not protect against every unexpected event.
  3. Check ordinary-stop execution and any guaranteed-stop eligibility, premium and restrictions. Record actual or assumed fills rather than a promised first available quote.
  4. Evaluate frozen rules on later data with costs, missed orders and drawdowns. Ten demo weekends, discipline or a 1% input does not establish future profit or live readiness.
  5. Use the rules of weekend trading and what a price gap is to check instrument availability and price definitions before designing the test.

The Upside Gap Three Methods formation and Downside Gap Three Methods compare particular body and gap relationships. They do not make every breakout gap a trend-following opportunity. Keep geometric recognition, timed signals and executed net results separate.

A gap changes the fill while the stop level stays fixed

About 2 min calculator practice

Use a separate EUR/USD long model: 0.12 lot, EUR 100,000 per lot, a USD account, entry fill 1.1000 and exit fill 1.0980. Price P/L is −USD 24, with no separate fees. Assume a worse close after a gap. Change only the closing execution price to 1.0965.

Calculate P/L from the worse exit fill

Check the result and explanation

The adverse price distance is 35 pips. P/L is 12,000 × (1.0965 − 1.1000) = −USD 42, a loss larger by USD 18. The calculator uses the assumed execution price; the extra 15 pips are not deducted again.

This is a separate hypothetical path, not a gap forecast. An ordinary stop does not guarantee its price; this calculation does not verify triggers or provider rules.