Do both island reversal gaps have to be the same size?

No. They must share an empty price interval under this page's rule. In the bearish example, the first gap is 3 units and the second is 2.5 units; their overlap is 2 units. Two gaps at unrelated price levels do not establish an isolated island.

Does an island reversal require a doji?

No. The island may contain one or more bars without a doji requirement. An abandoned baby is a narrower three-candle pattern with an isolated doji-like middle candle and additional surrounding candle conditions.

When is the island reversal confirmed?

This protocol confirms at the close of the exit bar when both full-range gaps and the shared empty interval still exist. At the open, the exit bar can appear separated but later trade into the island. A close-confirmed event does not establish a profitable reversal.

What invalidates an island reversal?

State the failure rule before testing. This page records a full exit-gap fill when a later high reaches the last island low for a top, or a later low reaches the last island high for a bottom. Partial fills and loss of the common empty band are recorded separately. The earlier confirmation remains in the log.

Is there a reliable island reversal success rate?

This page establishes no success rate. Any estimate needs a stated market, period, session, adjustment policy, detection rules and outcome definition. A strategy result also needs executable entries, exits and costs; selected chart examples cannot supply that estimate.

TradingView chart patterns

Island reversal pattern: two gaps and an isolated price area

An island reversal pattern has one or more price bars separated from surrounding trading by two gaps in opposite directions. An island top gaps up into the isolated area and down out of it after an advance. An island bottom reverses that sequence after a decline. Both gaps must leave a shared empty price interval; a different opening price alone is insufficient. The second gap completes the structure, but later price action can fill it and defeat the reversal interpretation.

See the decision guide

Define island tops and bottoms with the same range test

Arthur Hill describes the bullish sequence as a gap down, consolidation and a matching gap up. The following stricter measurement convention makes the isolated area testable. Let P be the bar immediately before the island, E the exit bar, and I all island bars between them. I-low is the lowest low in I; I-high is its highest high. All bars must be consecutive in the selected chart series.

The inequalities are strict: equal prices touch and leave no gap. A minimum gap in ticks can be added as an explicit filter, but it changes the sample.
Checking every island bar prevents an interior wick from crossing the supposed empty band. Two adjacent-bar gaps can exist without isolating the whole cluster.
For a repeatable context rule, this teaching example requires P-close above its close 20 bars earlier for a top, or below it for a bottom. That simple comparison is a chosen filter, not a complete trend definition or an optimized setting.

Check	Island top	Island bottom
Context	Prior advance	Prior decline
Entry gap	First island low > P-high	First island high < P-low
Exit gap	E-high < last island low	E-low > last island high
Shared empty interval	max(P-high, E-high) < I-low	I-high < min(P-low, E-low)
Time of confirmation here	Close of E, with all tests still true	Close of E, with all tests still true

Calculate a bearish island with three isolated bars

These invented daily ranges describe an island top after an advance. Assume P-close is 99, compared with 90 twenty bars earlier, so the stated context rule passes. Prices are in arbitrary units and do not represent a traded instrument.

Entry gap: the open interval (100, 103), width 3. Exit gap: (101, 103.5), width 2.5. Their common empty interval is (101, 103), width 2.
I-low is 103. The full-cluster test is max(100, 101) < 103, so every island bar stays above the shared empty interval.
The two gaps need not have identical endpoints or widths. They need a positive-width overlap under this rule.
Counterexample: if I2 had a low of 100.5, the two boundary gaps would still exist, but the whole cluster would fail the isolation test because 101 < 100.5 is false.

Bar	Low	High	Role
P	98	100	Last bar before the entry gap
I1	103	106	Gaps above P
I2	104	108	Remains inside the isolated cluster
I3	103.5	107	Final island bar
E	98	101	Gaps below I3; closes at 99

Mirror the calculation for a bullish island bottom

For a separate invented example, assume P-close is 101, below its close of 110 twenty bars earlier. P has a 100 to 102 range. The three island ranges are 94 to 97, 92 to 96, and 93 to 96.5. E trades from 99 to 102 and closes at 101.

Confirmation occurs at the close of E in this protocol. A later low at or below 96.5 fully fills the saved exit gap under the range-touch definition below.
A one-bar island can qualify without being a doji. TradingView defines an abandoned baby using an isolated doji-like middle candle and additional first/third candle requirements. A multi-bar island does not require that candlestick structure.
A gap-and-go setup has a different task: continuation after an opening gap. One gap and continued movement do not complete this two-gap reversal structure.

Measurement	Calculation	Result
Entry gap	(97, 100)	Width 3
Exit gap	(96.5, 99)	Width 2.5
Shared empty interval	(97, 99)	Width 2
Full-cluster isolation	97 < min(100, 99)	Passes

Separate partial fills, failed islands and assumed executions

Freeze the bearish example at E-close: the exit gap is (101, 103.5), and the common empty band ends at 103. This protocol calls the setup failed once a later bar trades at or above 103.5. It is a chosen range-touch failure rule; waiting for a close above that price would be a different test.

The rows are alternative next-bar paths. The final path is the quick-fill counterexample: a completed island can fail immediately. Do not delete its original signal from a test.
At an island top, the saved island high of 108 and the exit-gap edge of 103.5 are different possible risk references. Choosing either as an exit level changes the strategy; neither guarantees a fill at that price.
A bar that jumps across a gap is classified as filled by this range-touch rule even if no trade occurred at every intermediate price. OHLC bars alone do not establish the tick-by-tick path or order fills.
Define entry timing, order type, position size, holding period, costs and exit execution separately. A close-confirmed signal cannot retrospectively enter at the opening price of E.

Later observation	Classification	What to record
Next high = 102	Exit gap partly filled	1 of the original 2.5 units traversed; 40% fill
Next high = 103.2	Shared empty band lost; exit gap not fully filled	2.2 / 2.5 = 88% fill; 0.3 remains to the saved exit-gap edge
Next bar trades from 100 to 104 and closes at 103.6	Exit gap fully filled; reversal setup failed under this rule	Keep confirmation at E and failure on the next bar

Check sessions, adjustments and volume before counting signals

The chart series is part of the definition. TradingView allows extended-session data on supported intraday charts and dividend adjustments on supported assets. Changing those inputs changes the bars being compared; a visible empty area on one series need not survive on another.

Save the exchange-qualified symbol, data feed, ordinary candle type, timeframe, timezone, session and adjustment settings with each candidate. Inspect split and dividend dates before interpreting a discontinuity as trading pressure.
On intraday charts, compare regular and available extended hours to see whether omitted trading occupies the apparent gap. Do not silently mix their highs and lows in one test.
Compare adjusted and unadjusted prices where the platform supports it. This is a sensitivity check, not a reason to switch settings until a preferred pattern appears.
Volume is an optional condition, not a replacement for the two gaps. One declared experiment could require E-volume / mean volume of the previous 20 completed bars to be at least 1.5. For 1.8 million / 1 million, the ratio is 1.8 and passes. The threshold is hypothetical and has no measured advantage here.
Exclude E from its volume baseline. Missing volume or a zero baseline makes that filter unavailable. Keep the same session and feed for both numerator and denominator, then compare outcomes with and without the filter.
A full-range detector may find few events on a continuously traded series. Keep the same gap definition rather than substituting candle-body gaps or missing data to produce more signals.

Specify the detector before asking Pineify to draft it

Pine Script AI Coding Agent can help turn explicit conditions into code for review. Supply the high-low rules, trend filter, session assumptions, island-length bounds and failure rule. This page provides a specification and numerical checks; it does not scan symbols or provide a validated island detector.

For a bounded research version, consider 1 to 10 island bars and test only completed exit bars. If several lengths pass, report the longest and save its start/end timestamps. These are explicit example choices, not universal pattern rules.
Request separate candidate, confirmed and failed states. Keep the confirmation marker on E and preserve its original timestamp even if a later gap fill changes the current state.
Review generated code, request a syntax check, then compare signals against both numerical examples and the interior-wick counterexample. TradingView documents that current-bar highs and lows change during realtime updates; a provisional opening gap can disappear before close.
Use Trading Journal plan and review notes to save the intended entry, the actual fill, the session settings and any missed or rejected order. A 99 planned short entry filled at 98.5 is 0.5 units of adverse entry difference per unit; journal the observed prices instead of assuming execution at the signal.

Where Pineify fits

Pine Script AI Coding Agent

Draft a detector from your chosen gap, length and confirmation rules. Review its code, request syntax validation and compare its events with labeled charts before testing trades.

Also useful: Trading Journal. Record planned entries, actual trade prices and review notes about gap fills or missed executions. The notes do not automatically verify the chart pattern.

Draft my island detection rules

Related resources

TradingView chart patterns Abandoned baby candle requirements Bar-close alert timing Trading Journal

Frequently asked questions

Educational information, not investment advice. Prices, volume values and detection settings are hypothetical. No success rate, backtest or live performance is claimed. A completed island can fail, and gap trading can produce losses and fills far from intended prices.

Sources and verification

Last verified: 2026-10-17

Published by Pineify. Pineify is an independent product and is not affiliated with TradingView. Sources support the technical explanations; examples do not establish trading performance.