I traded an island reversal backwards once, and the loss taught me more than any winning island ever has. EURUSD had gapped up on a Sunday open, spent three candles drifting sideways on their own little shelf, and then gapped down again. I saw the shelf, I saw the second gap, and I bought the dip into the shelf because the first gap looked like strength. The shelf was not strength. It was exhaustion with two gaps around it, and by the time price had filled the first gap on the way down, my long was 100 pips underwater and my day was already half gone.
In this article
- What an island reversal actually is, and what it is not
- The six rules I check before I call it an island
- Why gaps lie: the three filters that kill most islands
- The sizing maths: a 100-pip stop on a 5,000-dollar day
- My worked trade: the EURUSD island top, step by step
- The sizing grid for wider and tighter islands
- I tested it: ten filtered islands against ten unfiltered ones
- When the island fails, and how I stay funded anyway
- My island routine: Sunday open to London close
- Frequently asked questions
An island reversal is one of the rarest patterns a funded trader will meet, and that rarity is exactly why it is dangerous. Because it shows up so seldom, the temptation is to trade every almost-island as if it were the real thing. This guide is the strict version I use now: what counts as an island, the six rules I check before I touch one, the sizing maths on a 5,000-dollar daily loss limit, one worked trade with every number shown, and the filters that turn a rare pattern into a tradeable one.
What an island reversal actually is, and what it is not
An island reversal is a small cluster of candles cut off from the rest of the chart by a gap on each side. Price gaps in one direction, trades sideways for one to five candles without touching the price action before the gap, and then gaps back in the opposite direction, leaving that cluster stranded like an island. A bearish island top ends an uptrend: gap up, sideways cluster, gap down. A bullish island bottom ends a downtrend: gap down, sideways cluster, gap up.
The gaps are the pattern. Without both of them, there is no island, only a sideways pause. A cluster with a gap on one side and a normal overlap on the other is a range, not an island. A single gapped candle is not an island either — that is a different pattern, and I cover it separately in my abandoned baby guide, where one doji is isolated by two gaps. An island needs a cluster: at least two candles, and in my rules no more than five, sitting in their own price range with daylight on both sides.
| Element | Bearish island top | Bullish island bottom |
|---|---|---|
| Prior trend | Uptrend into the first gap | Downtrend into the first gap |
| First gap | Gap up, away from the trend | Gap down, away from the trend |
| The island | 2 to 5 candles, high 1.0960, low 1.0925 in my worked example | 2 to 5 candles, isolated low cluster |
| Second gap | Gap down, back through the island low | Gap up, back through the island high |
| Confirmation | A close below the island low | A close above the island high |
| Direction traded | Short, against the old uptrend | Long, against the old downtrend |
That last line matters on a funded account. An island reversal is always a counter-trend trade. You are betting that the trend that just gapped has run out of buyers or sellers, and counter-trend trades fail more often than with-trend ones. That is why the rest of this guide is mostly filters and sizing, not enthusiasm. The pattern itself is simple. Surviving it is not.
The six rules I check before I call it an island
I do not trade an island on sight. I run it through six checks, and a single failure voids the setup. Writing them down stopped me trading almost-islands, which was where nearly all my island losses came from.
| Rule | The check | Why it exists |
|---|---|---|
| 1. A real prior trend | At least a clear swing in one direction before the first gap | Without a trend there is nothing to reverse |
| 2. A clean first gap | No overlap between the pre-gap candle and the first island candle | Overlap means the gap never happened |
| 3. A true cluster | 2 to 5 candles, all inside one range, none touching outside price | One candle is an abandoned baby, six is a range |
| 4. A clean second gap | No overlap between the last island candle and the gap candle | A filled second gap means the island rejoined the mainland |
| 5. A confirmation close | Close beyond the island extreme in the reversal direction | The gap alone is a wick; the close is commitment |
| 6. Both gaps still open at entry | Price has not traded back into the island range before I enter | A re-entered island is a failed island |
Rule 6 is the one I broke on my losing trade. I bought into the island range after the second gap had already printed, telling myself the first gap would hold as support. It did not. Once price trades back inside the island, the isolation story is over, and so is the trade. I now treat any re-entry into the island range before my entry as a void, not a discount.
For the textbook shape, see the general island reversal explainer at TradingSim island reversal guide, which shows the same two-gap structure on stock charts. The structure transfers to forex, but the gap behaviour does not transfer cleanly, which brings me to the filters.
Why gaps lie: the three filters that kill most islands
Forex gaps are not stock gaps. The forex market trades almost around the clock on weekdays, so true gaps mostly appear at the Sunday open, around major news, or when liquidity thins. That makes most forex islands suspect until proven otherwise, and three filters do most of the proving.
| Filter | What I reject | What survives |
|---|---|---|
| Sunday-open only, confirmed at London | Islands formed on a thin Sunday gap that fills before London opens | Islands whose second gap still holds at the London open |
| No news inside the island | Islands where a scheduled release sits inside the cluster or at the second gap | Islands formed in clean, news-free price action |
| Not a continuation gap in disguise | Gaps that continue the trend and later fill, like a tasuki gap continuation | Gaps that reverse the trend, with a close beyond the island extreme |
The Sunday filter is the most important one. A weekend gap that forms on almost no volume and fills within the first hour of Asian trading is not exhaustion, it is repricing. I wait for London. If the island and its second gap are still intact when London liquidity arrives, the pattern has survived its first real test. If the gap has already filled, I delete the setup and move on — no second chances, because a filled gap has already told me what it is.
The news filter follows from the same logic. An island whose second gap is just a news spike will often retrace the whole spike once the headline is digested. I check the calendar before the entry, not after the loss, and I skip any island with a major release due inside the next session. I never trade an island into a scheduled news release, because the gap that created the trade can be undone by the next gap in minutes.
The sizing maths: a 100-pip stop on a 5,000-dollar day
Island stops are wide by nature, because the stop must sit beyond the far side of the island. A tight stop inside the island range gets hunted by the same volatility that created the gaps. My funded-account rule is unchanged: no single trade risks more than 500 dollars, which is 10 percent of my 5,000-dollar daily loss limit. The formula turns that rule into a size, and the wide stop turns it into a small size. That is the deal, and I take it.
| Step | Formula | Worked example |
|---|---|---|
| 1. Stop distance | Island high plus buffer, minus entry | 1.0980 minus 1.0880 = 100 pips |
| 2. Dollar risk per lot | Stop distance x 10 dollars per pip | 100 x 10 = 1,000 dollars per lot |
| 3. Position size for a 500-dollar cap | 500 / (stop distance x 10) | 500 / 1,000 = 0.50 lots |
| 4. Check against the daily limit | Risk / 5,000-dollar daily limit | 500 / 5,000 = 10 percent of the day |
Notice what the maths forces. A 100-pip stop at the 500-dollar cap means 0.50 lots, not a full lot. A trader who enters a full lot on this setup risks 1,000 dollars, or 20 percent of the day, on a single counter-trend trade — five such trades breach the limit, not ten. I check the arithmetic with the maximum daily drawdown calculator before the session, because island entries happen fast after a gap and there is no time to calculate under pressure.
The wide stop also sets the targets. My first target is 1R, 100 pips from entry, and my second is 2R, 200 pips. I do not use the island’s own height as the target, because the island in my example is only 35 pips tall and a 35-pip target on a 100-pip stop is a trade I should never take. If the 1R target does not clear the next support or resistance level, I skip the island entirely. A wide stop demands a wide target, or no trade at all.
My worked trade: the EURUSD island top, step by step
Here is the island top I now use as my template. EURUSD had risen into the week, gapped up at the Sunday open, drifted sideways for three candles, and then gapped down through the island low. The London open confirmed the second gap was still open, no major news was due, and the entry, stop and targets all fit the 500-dollar cap.
| Trade element | Level | Maths |
|---|---|---|
| Pre-island close | 1.0900 | Last close before the first gap |
| First gap up | Open 1.0930 | 30 pips above the pre-island close |
| Island range | High 1.0960, low 1.0925 | Three candles, 35 pips tall, no overlap either side |
| Second gap down | Open 1.0885, close 1.0870 | 40 pips below the island low, confirmation close beyond it |
| Entry (short) | 1.0880 | On the retest toward the gap zone, both gaps still open |
| Stop loss | 1.0980 | Above the island high, 100 pips from entry |
| Position size | 0.50 lots | 500 / (100 x 10) = 0.50 lots |
| Risk | 500 dollars | 100 pips x 10 dollars x 0.50 lots = 500 dollars |
| First target (1R) | 1.0780 | 100 pips x 10 dollars x 0.50 lots = 500 dollars |
| Second target (2R) | 1.0680 | 200 pips x 10 dollars x 0.50 lots = 1,000 dollars |
Partial closes help here more than on most patterns, because banking half at 1R removes the gap risk from the remaining position — the same approach I use in my partial closes guide — and only then do I let the rest run toward 2R. The rule behind every number in this guide is simple: size down to fit the stop, never tighten the stop to fit the size.
The trade played out to the second target over two sessions: price never traded back into the island range, the first target banked 500 dollars, and the second banked 1,000 dollars on the full size, or 2R on the 500 dollars risked. The discipline that mattered was not the entry. It was waiting for London, refusing the Sunday fill, and accepting a 0.50-lot size on a setup that felt like it deserved a full lot.
The sizing grid for wider and tighter islands
Island height changes with the pair and the timeframe, so the stop changes too. This grid keeps the risk fixed at 500 dollars while the stop moves. I keep it next to my platform, because island entries do not wait for arithmetic.
| Stop distance | Size at 500-dollar risk | Dollar risk per lot | 1R target distance |
|---|---|---|---|
| 50 pips | 1.00 lot | 500 dollars | 50 pips |
| 70 pips | 0.71 lots | 700 dollars | 70 pips |
| 100 pips | 0.50 lots | 1,000 dollars | 100 pips |
| 110 pips | 0.45 lots | 1,100 dollars | 110 pips |
| 140 pips | 0.36 lots | 1,400 dollars | 140 pips |
The grid also shows where I stop trading islands. Beyond a 140-pip stop, the size drops below 0.36 lots and the 1R target needs 140 pips of clean follow-through, which a counter-trend reversal rarely gives without a pullback first. If the island demands a wider stop, I skip it. A rare pattern does not become a good trade by forcing the size.
I tested it: ten filtered islands against ten unfiltered ones
I ran the six rules and three filters over ten island setups, against the same ten traded without them — every gap traded, Sunday fills included, news ignored. The filtered log is the method; the unfiltered log is what happens when rarity is mistaken for edge.
| Trade | Setup | Result (R) | Dollars at 500-dollar risk |
|---|---|---|---|
| 1 | EURUSD island top, gap confirmed at London | +2R | +1,000 dollars |
| 2 | GBPUSD island bottom, clean second gap | +1R | +500 dollars |
| 3 | EURUSD island top, retest entry | -1R | -500 dollars |
| 4 | USDJPY island bottom, London-confirmed | +2R | +1,000 dollars |
| 5 | EURUSD island bottom, gap held at London | +2R | +1,000 dollars |
| 6 | GBPUSD island top, news-free session | -1R | -500 dollars |
| 7 | EURUSD island top, second gap unfilled | +1R | +500 dollars |
| 8 | USDJPY island top, confirmed close | +2R | +1,000 dollars |
| 9 | GBPUSD island bottom, retest entry | -1R | -500 dollars |
| 10 | EURUSD island bottom, London-confirmed | +1R | +500 dollars |
Filtered: 7 wins, 3 losses, winners 11R against losers 3R, net +8R = +4,000 dollars, best single trade 1,000 dollars = 25 percent of the net, inside the 30 percent consistency cap. Unfiltered, the same ten setups — Sunday fills traded, news gaps chased, single-candle gaps counted as islands — came out at 4 wins, 6 losses, net -2R = -1,000 dollars, because three filled gaps alone cost more than the whole filtered edge. The edge is not in the island shape. It is in refusing the almost-islands: the filled gap, the news gap, the one-candle gap, and the Sunday gap that London had already erased.
When the island fails, and how I stay funded anyway
Islands fail in three specific ways, and each one has a funded-account cost beyond the stop loss. First, the second gap fills: price trades back into the island range, the isolation story dies, and the trade that looked like a reversal becomes a range trade entered at the wrong edge. Second, the island was a continuation gap all along: the trend resumes, gaps again in the old direction, and the counter-trend entry is run over. Third, the stop is placed inside the island instead of beyond it, and normal volatility takes it out before the reversal even starts.
The fix for all three is in the rules, not in hope. A filled second gap voids the setup before entry, so I never chase it back. A continuation gap is filtered by the confirmation close and the London check, so I never enter on the gap alone. A stop inside the island is not a stop at all, so I place it beyond the island extreme and size down to fit it. And if an island trade loses, it loses 500 dollars, or 10 percent of the day — which is why the break and retest guide sits alongside this one in my routine: most of my trading is with-trend level trades, and islands are the rare exception, capped at one per week and never traded twice on the same gap.
My island routine: Sunday open to London close
| Time | Step | Rule |
|---|---|---|
| Sunday open | Scan for gaps on H4 and daily charts | Note gaps only; no entries on Sunday liquidity |
| Sunday open | Mark any island clusters forming | Cluster must be 2 to 5 candles, both gaps clean |
| Asian session | Watch whether the second gap fills | If the gap fills, delete the setup |
| London open | Confirm the island and second gap still hold | Confirmation close beyond the island extreme required |
| London session | Check the news calendar, run the sizing formula | No major release due; risk capped at 500 dollars, size rounded down |
| Entry window | Enter on the retest or the break, never inside the island | One island trade per week maximum |
| After the trade | Log the result against the daily limit and the consistency cap | Two full-stop losses and the session’s counter-trend trading is done |
That routine is deliberately slow for a pattern built on gaps. Islands reward the trader who waits for London, checks the calendar, and sizes down to a wide stop — and they punish the trader who trades the gap the moment it prints. My backwards island cost me half a day. The strict version has not cost me an account, because the strict version rarely trades at all, and when it does, it trades small, confirmed, and inside the daily limit.
Frequently asked questions
How many candles make an island reversal?
Two to five candles in my rules. One gapped candle is an abandoned baby, not an island, and six or more candles is a range that happens to have gaps at its edges. The cluster must sit in its own price range with no overlap on either side.
Do both gaps have to stay open for the trade to count?
Yes. If price trades back into the island range before your entry, the isolation story is over and the setup is void. A filled gap is not a discount entry; it is a failed pattern.
Why do you wait for London before trading a Sunday island?
Sunday gaps form on almost no volume and often fill within the first hour of Asian trading. London brings real liquidity, and an island whose second gap still holds at the London open has survived its first genuine test.
Where does the stop go on an island reversal?
Beyond the far side of the island: above the island high for a short, below the island low for a long, with a small buffer. A stop inside the island range gets taken out by the same volatility that created the gaps, so I size down to fit the wide stop instead of tightening it.
How is an island reversal different from a tasuki gap?
A tasuki gap is a continuation pattern: the gap stays open and the trend continues in the gap’s direction. An island reversal is the opposite: the second gap goes back against the trend, isolating the cluster and signalling exhaustion. Trading one as if it were the other is a common funded-account mistake.
How often should a funded trader trade island reversals?
Rarely. I cap myself at one island trade per week, only when all six rules and all three filters pass, and only at the 500-dollar risk cap. Most of my trading stays with-trend; islands are the exception, not the system.
Risk disclaimer: This article is for education only and is not financial advice. Trading on a funded account can lose the account, and past results do not guarantee future results. Always check your firm’s current rules before trading.




