Most reversal patterns describe a turn. The island reversal describes a stranding. Price gaps away from where it was trading, spends a while somewhere new, and then gaps back the way it came, leaving a cluster of candles sitting alone on the chart with empty space on both sides. Everyone who bought or sold inside that cluster did so at prices the market has since abandoned in both directions.
An island reversal pattern is a cluster of candles cut off from the rest of the chart by two gaps at roughly the same price: one in the direction of the prior trend, one against it. An island top follows an advance, an island bottom a decline. The read is invalidated by a close back through the second gap.
It is one of the most visually striking patterns on a chart, and one of the most over-described. The measured record is not kind to it, and the pattern only exists at all in markets that gap. Here is what it requires and where it misleads.
What does an island reversal pattern need, in checkable terms?
Bulkowski’s identification guidelines are short and strict:
- A prior trend into the island. Island tops have price trending upward into the island; island bottoms have price trending downward. Without it there is nothing to reverse.
- Two gaps. A gap is a price range with no trading at all, where one candle’s low is above the previous candle’s high (a gap up) or one candle’s high is below the previous candle’s low (a gap down). The first gap carries price onto the island; the second carries it off.
- The gaps share a price. The two gaps must overlap, sharing some or all of the same price range. This is what makes it an island rather than two unrelated gaps. Bulkowski separately describes a “long island” where the gaps do not align, and treats it as a different and, as of 2020, no longer valid pattern.
- Any duration. The island can be a single candle or several months of trading.
He also lists heavy volume on the day of the second gap as a characteristic. If the volume tradition is new, the volume post covers what a tall bar on a gap day does and does not measure.
An illustrative island top, with prices made up for the purpose: a stock trends up from 30 over six weeks, and on the last day of the advance its high is 43.80. The next morning it gaps up, and that session’s low is 44.20, leaving a gap from 43.80 to 44.20. For eight sessions it trades between 44.20 and 47. Then it gaps down: the next session’s high is 43.60, below the island’s low of 44.20, leaving a second gap from 44.20 down to 43.60. The two gaps share the range from 43.80 to 44.20: that is the island.
What does the island actually record?
Each gap is a moment when the price jumped past a range where nobody traded, usually because news arrived while the market was closed. The first gap says buyers (for a top) were willing to pay well above the prior close to get in. The second gap says that, some time later, the same market was willing to sell well below the island’s lowest price to get out.
The participants trapped on the island are the mechanism the tradition points to. Everyone who bought during the island is now holding a position below their entry, with no trading between their price and the current one. Whether that pressure actually produces a continued move is what the measured data tests.
Confirmation and invalidation
- Needs: a trend into the island, and two gaps that overlap in price.
- Confirmed by: the second gap itself. It completes the pattern and is the breakout; until it prints, there is a gap and a cluster of prices, not an island.
- Invalidated by: price trading back into the island and closing beyond the second gap, which fills it. For an island top, a close back above the island’s low; for an island bottom, a close back below the island’s high. Once the gap is filled, the island is no longer stranded.
This is the “gap fill” that so many searches ask about. Gaps do often fill, which is exactly why an island is fragile. Bulkowski’s figures put the pullback rate for island tops at 55% and the throwback rate for island bottoms at 54%, meaning that in roughly half his cases price returned to the breakout price within 30 days. A return to the gap is common; a close back through it is the line that ends the read.
What does the measured record say?
Bulkowski’s bull-market statistics for islands, updated in August 2020 and based on what he describes as over 2,000 perfect trades (the page does not state the market or the period the sample covers), are unusually blunt:
- Performance rank: island bottoms rank 38th of 39 chart patterns for upward breakouts, and island tops 31st of 36 for downward breakouts.
- Break-even failure rate: 31% for upward breakouts and 34% for downward breakouts, where his glossary defines a break-even failure as a move of less than 5% in the breakout direction.
- Average move: a rise of 31% and a decline of 13%, measured from the open the day after the breakout to the highest peak or lowest valley before a 20% reversal, which is a generous measure.
Beneath his own worked example, he writes that you can see why islands do not work well. That is the honest summary from the analyst who has measured them most closely. The pattern is dramatic, rare and memorable, and those three qualities are not the same as reliable.
These are one analyst’s counts on his own sample, and they do not transfer to another market or timeframe. They are enough to say that the island’s reputation as a powerful reversal is not supported by the most detailed public data.
Why do islands barely exist in some markets?
An island needs two gaps, and gaps need a market that stops trading. US stocks have a natural one every night: the regular session ends at 4 p.m. Eastern and the next one opens at 9:30 a.m., and news arrives in between. That overnight break is where most stock gaps come from.
Markets that trade continuously remove most of those breaks. On a currency pair, a new candle opens essentially where the last one closed between Monday and Friday, so gaps occur almost only over the weekend; the forex candlestick patterns post covers what that does to gap-based definitions. On crypto, which never closes, a true gap on a liquid pair is rare. What looks like an island on those charts is usually a cluster of candles separated by a fast move rather than by a range where no trades occurred, and it is a different, weaker observation.
On intraday stock charts, the same rule applies in reverse: the only real gap on a 5-minute chart is usually the overnight one at the open. An “island” that forms inside a single session was not separated by gaps in the sense the pattern means.
Where it misleads
Gaps that do not overlap. Two gaps in opposite directions at different prices are two gaps, not an island. The overlap requirement is what makes the stranding real.
No trend into it. An island that forms in the middle of a sideways range has nothing to reverse. It is two gaps inside a range, which happen for news reasons and usually say little about direction.
The second gap fills within days. This is the most common way an island fails. The invalidation exists precisely for this case, and the pullback rate above says it happens often.
Treating a dramatic shape as strong evidence. The visual isolation makes islands feel significant. The measured ranks say otherwise. A pattern is worth what its record says, not what it looks like.
What this post is not
It is not a reason to look for islands. It is a way to describe one accurately when you see it: a trend in, two overlapping gaps, a cluster of trapped prices, and a specific close (back through the second gap) that ends the read. The triple top post applies the same discipline to a far more common reversal shape, and how to read a stock chart covers the gaps and sessions that make patterns like this possible in the first place.
Sources
Every figure above was checked against these before publishing.
- Bulkowski, Island Tops and Island Bottoms
- Bulkowski, Long Islands
- Bulkowski, Glossary
- NYSE, Hours and Calendars