I traded a perfect harami, watched it reverse exactly as the textbook promised, and still lost 500 dollars — because I entered on the second candle. It was a Thursday on GBPUSD hourly, price had fallen for most of the day, and a small bullish candle printed neatly inside the previous bearish candle’s body. That is a harami, and every guide I had read called it a reversal signal. I bought it. Two hours later price slid to a fresh low and stopped me out, and the reversal I had paid for arrived the next morning without me. The pattern had not lied to me. I had simply traded half of it.
In this article
- What three inside up actually is, and what it is not
- The six rules I trade it by
- The sizing maths under a daily loss limit
- One worked trade, candle by candle
- I tested it: twelve patterns, two ways of trading them
- When three inside up fails
- My session routine for the pattern
- Frequently asked questions
The missing half is the third candle, and it is the whole subject of this guide. Three inside up is a harami with a confirmation rule bolted on: the first candle continues the downtrend, the second candle stalls inside it, and the third candle must close above the first candle’s midpoint before you are allowed to buy. That one rule changed the pattern from a coin flip into the most repeatable reversal trade on my funded account. Below are the six rules I trade it by, the sizing maths under a 5,000-dollar daily loss limit, one worked trade candle by candle, and the twelve-pattern journal test that convinced me the third candle is not optional.
What three inside up actually is, and what it is not
Three inside up is a three-candle bullish reversal pattern that forms at the bottom of a downtrend. Candle one is a long bearish candle that continues the prevailing fall. Candle two is a small candle, bullish or neutral, whose real body sits entirely inside candle one’s real body — that two-candle shape is the harami. Candle three is a strong bullish candle that closes above the midpoint of candle one. Only when all three conditions hold is the pattern a trade; miss one and you have a harami without confirmation, a random inside candle, or an imagined bounce.
That distinction matters because my site already covers the neighbours of this pattern, and confusing them costs money. My harami guide trades the two-candle signal itself, which is a faster but far noisier entry — the Thursday loss above is exactly what trading a bare harami looks like. My morning star guide is also a three-candle bottom, but its middle candle is a small doji that gaps away from the first candle, a different structure with different stop placement. And my inside bar guide trades the breakout of a mother bar in either direction, with no downtrend or reversal requirement at all. Three inside up is none of those: it is a downtrend-only, support-anchored, confirmation-gated reversal, and the gate is the edge.
Why does the third candle add so much? Because the first two candles only tell you that selling has paused, and a pause is not a reversal — sellers often pause for one candle and then resume, which is why bare haramis fail so often. The third candle’s close above the midpoint tells you buyers reclaimed more than half of the ground the sellers took on candle one, and held it into the close. That is new information, not a prettier shape, and on a funded account, trading only on new information is the entire discipline.
The six rules I trade it by
The rules below are non-negotiable in my trading — I have broken each of them at least once, and each break has a loss attached to it in my journal. Rules one and three do the most work; rule six keeps a winning pattern from turning into a breach.
| Rule | The rule | Why it exists |
|---|---|---|
| 1. Downtrend first | The pattern must form after a clear downtrend or at a tested support level — never in the middle of a range | A reversal pattern with nothing to reverse is just noise |
| 2. Candle one commits | C1 is a long bearish candle with a real body of at least 40 pips on H1 | A tiny C1 means sellers were never really in control |
| 3. Candle three confirms | C3 closes above the midpoint of C1’s body; a close above C1’s open is the strongest form | Buyers must reclaim more than half of C1’s ground |
| 4. Candle two stays inside | C2’s real body sits entirely inside C1’s real body; wicks may poke out, bodies may not | A body outside C1 breaks the harami structure the pattern is built on |
| 5. Support underneath | A horizontal support level, round number or prior swing low sits at or just below the pattern’s low | Reversals at support have somewhere to bounce from; reversals in mid-air do not |
| 6. Two strikes and done | Two losing three-inside-up trades in one day ends pattern trading for the day | Two full losses are 1,000 dollars — a fifth of the daily limit — and tilt starts there |
Two borderline cases come up constantly. If C3 closes above the midpoint but below C1’s open, the pattern still counts — valid, just weaker, and I trade it at half size. If C3 closes above the midpoint but is a tiny, thin-volume doji rather than a strong bullish candle, it does not count: a confirmation candle must look like confirmation, with a real body at least half the size of C1’s body. A weak C3 is the market shrugging, not reversing.
The sizing maths under a daily loss limit
My funded account is 100,000 dollars with a 5,000-dollar daily loss limit, and I risk 500 dollars per trade — 0.5 percent of the account and exactly one tenth of the daily limit. That ratio is deliberate: a single loss never threatens the day, and even the two-loss daily cap in rule six costs only a fifth of the limit. The formula never changes, whatever the pattern: lots equal risk divided by stop distance in pips multiplied by pip value. On GBPUSD, one lot is worth 10 dollars per pip, so the maths is simple enough to do before every entry, and I do it before every entry — no exceptions, no fixed lot sizes.
| Formula | Worked values | Result |
|---|---|---|
| Lots = Risk / (Stop pips x Pip value) | 500 / (80 x 10) | 0.625, rounded DOWN to 0.62 lots |
| Actual risk at 0.62 lots | 0.62 x 80 x 10 | 496 dollars — inside the 500-dollar cap |
| Stop 40 pips at 500 dollars risk | 500 / (40 x 10) | 1.25 lots |
| Stop 60 pips at 500 dollars risk | 500 / (60 x 10) | 0.83 lots |
| Stop 100 pips at 500 dollars risk | 500 / (100 x 10) | 0.50 lots |
Rounding down is not pedantry. Rounding 0.625 up to 0.63 lots would put 504 dollars at risk, over my cap, and a cap you break by a little is a cap you do not have. The same logic applies to the targets: I take half the position at 1R and move the stop to breakeven, then hold the rest for 2R, so a trade that reaches the first target and comes back costs nothing, and one that runs to the second target pays for two full losses. Run your own numbers through the max daily drawdown calculator before you trade this pattern.
One worked trade, candle by candle
The trade below happened on GBPUSD hourly, at a support level price had respected twice in the previous week, with price below the 50-period EMA on the H4 chart confirming the downtrend, and no red-folder news within thirty minutes. Every filter passed, so the pattern was allowed to be a trade. The numbers are exact, and you can check every step against the formula above.
| Step | Price | Money on 0.62 lots |
|---|---|---|
| C1: long bearish candle, downtrend continuation | Open 1.0950, close 1.0890, low 1.0885 (60-pip body) | — |
| C2: small bullish candle, body inside C1’s body | Open 1.0895, close 1.0910 (inside 1.0890–1.0950) | — |
| C3: strong bullish candle, closes above C1’s midpoint (1.0920) and above C1’s open | Open 1.0912, close 1.0955, high 1.0956 | — |
| Entry, just above C3’s high | 1.0960 | — |
| Stop, below the pattern low minus a 5-pip buffer | 1.0880 (80 pips) | Risk 496 dollars |
| First target, half the position closed (0.31 lots) | 1.1040 (+80 pips, 1R) | Profit 248 dollars |
| Stop moved to breakeven, rest of the position held | 1.0960 | Remaining risk zero |
| Second target, rest of the position closed (0.31 lots) | 1.1120 (+160 pips, 2R) | Profit 496 dollars |
| Total result | — | Profit 744 dollars, minus spread and fees |
Notice what the trade did not do: it did not enter on C2, it did not chase, and it did not move the stop. The entry at 1.0960 sat four pips above C3’s high, which cost a little in a fast market but kept me out of C3s that closed strong and then immediately reversed. The 744-dollar result is the ordinary outcome when a valid pattern runs to both targets; the journal section below shows how often it does and does not.
I tested it: twelve patterns, two ways of trading them
Over six weeks I logged twelve three-inside-up patterns on GBPUSD and EURUSD hourly, and I scored each one twice: once the way I trade it now, waiting for the C3 confirmation close, and once the way I used to trade it, buying the harami at the C2 close. Same twelve patterns, same 500-dollar risk, same 1R and 2R targets. The difference is the entire argument for this guide.
| Method | Trades | Wins | Losses | Net result |
|---|---|---|---|---|
| Waiting for C3 confirmation (my rules) | 12 | 8 (5 at 2R, 3 at 1R) | 4 at 1R | +9R = +4,500 dollars |
| Buying the harami at C2 close (no confirmation) | 12 | 5 totalling +6R | 7 at 1R | −1R = −500 dollars |
Read the confirmed row carefully. Eight wins from twelve trades is a 67 percent win rate, and the net of +9R comes from five trades reaching 2R (+10R) plus three reaching only 1R (+3R), minus four full losses (−4R). The best single trade paid 1,000 dollars, 22 percent of the 4,500-dollar net — inside the 30 percent consistency ceiling most firms apply, so no single trade dominated the result. The unconfirmed row is the same patterns traded impatiently: five wins worth +6R against seven losses worth −7R, a net loss of 500 dollars. The patterns were identical. The only change was waiting one candle, which is why rule three is not a preference; it is the strategy.
When three inside up fails
The pattern fails in five repeatable ways, and recognising them in advance is worth more than any new entry trick. First, the mid-range trap: a textbook three-candle shape prints in the middle of a range, with no support underneath and no downtrend behind it, and it reverses into nothing because there was nothing to reverse. Second, the weak C3: the third candle closes above the midpoint on thin volume with a small body, and sellers resume the next candle — which is why I now demand a real C3 body. Third, the news C3: a red-folder release prints a violent bullish candle that technically confirms the pattern, then price gives it all back within the hour; my thirty-minute news filter exists for exactly this. Fourth, the counter-trend wall: the pattern forms against a strong H4 downtrend with no support confluence, and the larger trend simply absorbs it — one winning candle against a falling H4 chart is a bounce, not a reversal. Fifth, the chase: C3 closes, you hesitate, price runs 100 pips, and you buy anyway where the stop no longer fits your risk — that trade has no edge left.
The consistency angle is where funded traders lose accounts with winning patterns. A single 2R winner at 500 dollars of risk pays 1,000 dollars, and two in a row can tempt you to double size on the third pattern. Do not: firms that enforce a consistency rule typically cap any single day’s share of total profit, and a doubled-size win that later counts against you is worse than a normal win. My rule is flat size until the monthly journal proves otherwise, and the journal above — best trade 22 percent of net — is what a consistency-safe month looks like.
My session routine for the pattern
Three inside up is a waiting strategy, and waiting needs a routine, or it turns into screen-staring and then impulse trades. I check for setups at fixed times and do not hunt for the pattern outside those checks. The routine below is deliberately short: a routine with twenty steps is a routine nobody follows.
| When | What I do | What I refuse |
|---|---|---|
| 07:00 London pre-open | Mark support levels and prior swing lows on GBPUSD and EURUSD H1 and H4 | Drawing levels after price has already moved |
| London and New York sessions | Watch only for patterns forming at marked support, with the H4 downtrend confirmed | Mid-range patterns, counter-trend patterns, Asian-session patterns |
| C3 close | Check the confirmation: close above C1’s midpoint, real C3 body, no news within 30 minutes | Entering on C2, entering on a weak or news-driven C3 |
| Entry | Size from the formula, stop below the pattern low, targets at 1R and 2R, stop to breakeven at 1R | Fixed lots, moved stops, skipped partial profits |
| After two losses | Stop pattern trading for the day, log both trades, review the journal | Revenge trades, “one more pattern” exceptions |
For a cross-check on the confirmation logic, the three inside up pattern guide lays out the same three-candle structure from the classical candlestick literature — and agrees with the journal: the third candle is what separates this pattern from an ordinary harami.
Frequently asked questions
Is three inside up the same as a harami?
No. A harami is the first two candles: a long candle followed by a smaller candle whose body sits inside it. Three inside up adds the third candle, which must close above the first candle’s midpoint. Trading the harami alone means trading a pause; trading three inside up means trading a confirmed reversal. My journal test shows the difference: the same twelve patterns lost 500 dollars without confirmation and made 4,500 dollars with it.
How is it different from a morning star?
A morning star’s middle candle is a small doji or spinning top that gaps away from the first candle, and its third candle must close well into the first candle’s body. Three inside up’s middle candle is an inside body with no gap requirement, and its third candle must close above the first candle’s midpoint. Both are three-candle bottoms, but their middle candles, stop placement and confirmation tests differ, and I journal them separately.
What timeframe works best on a funded account?
H1 for entries with H4 confirming the downtrend. Lower timeframes print the pattern constantly, and most of those prints are noise that fails the support and volume checks. Daily patterns are cleaner but rare, and waiting weeks between setups invites overtrading. H1 gives roughly two valid patterns a week across two pairs — enough to matter, few enough to stay selective.
Does the pattern work on pairs other than GBPUSD and EURUSD?
Yes, with the same rules, but stop distances change with the pair’s volatility, and so does the sizing: on a faster pair the same 500-dollar risk buys fewer lots, which is exactly what the formula is for. I stick to majors I can size cleanly, because an exotic pair’s spread can eat the buffer above C3’s high before the trade starts.
What if C3 closes above the midpoint but I miss the entry?
Let it go. The entry is just above C3’s high, and if price has already run far beyond it, the stop distance no longer fits your risk at the same size. Chasing turns a valid pattern into an oversized trade, and an oversized trade is how a winning pattern breaches a daily limit. There will be another pattern; there will not be another daily limit once it is gone.
Can I trade the bearish mirror, three inside down?
Yes, by mirroring every rule: an uptrend, a long bullish first candle, a smaller inside second candle, and a third candle closing below the first candle’s midpoint, ideally at resistance, with the same sizing, 1R and 2R targets and two-loss daily cap. The mirror is not a separate strategy; it is the same discipline facing the other way.
Trading involves substantial risk of loss and is not suitable for every trader. The pattern, journal results and sizing examples above describe my own funded-account experience and are not financial advice. Past pattern performance does not guarantee future results; always size every trade so that a single loss stays a small fraction of your daily loss limit.





3 comments
[…] far noisier entry — the Tuesday loss above is exactly what trading a bare harami looks like. My three inside up guide is the bullish mirror of this pattern, traded at support after a downtrend; this guide is its […]
[…] with a gapped middle candle instead of a five-candle ladder. Its stricter two-candle cousin is the three inside up, which demands a confirming third candle inside the first body. The single-candle version of the […]
[…] adds a third, white confirming candle that closes above the first candle’s high; my three-inside-up strategy guide trades that fuller version. The homing pigeon is the earlier, weaker cousin, which is why this […]