I shorted an evening star in March 2025 and watched it do everything right except make me money. The pair was GBPUSD, the pattern sat at the top of a clean uptrend, and the third candle closed hard into the first candle’s body. I sold the next morning feeling clever, price drifted sideways for two days, and then it ground upward through my stop while I told myself the pattern was still valid. It was not. What I had traded was a shape, not a setup: no resistance level behind it, no confirmation candle, and a stop placed where I felt comfortable rather than where the pattern was actually wrong. That 480-dollar loss taught me that on a funded account a reversal pattern is only worth trading when every part of it is confirmed before entry, because a failed reversal does not just lose money, it hands the trend fresh fuel and leaves you short into strength.
In this article
- What an evening star actually is, and what it is not
- The anatomy checklist I run before I even think about selling
- The full strategy rules: setup, entry, stop, target and session filter
- Position-sizing maths under a 5,000-dollar daily loss limit
- One worked trade, candle by candle
- I tested it: twelve filtered trades against the unfiltered version
- When the evening star fails, and how it fits the funded rules
- Frequently asked questions
The evening star is the mirror image of the morning star. Where the morning star marks buyers taking control at a low, the evening star marks sellers taking control at a high: a long bullish candle, a small star candle that gaps up and stalls, and a long bearish candle closing deep into the first candle’s body. My rule is that the pattern only counts at a level that already matters, when the third candle closes below the midpoint of the first candle, and when one more candle confirms before I sell. This guide gives you the full rules I trade now, the sizing maths that keep each loss inside a 500-dollar slice of a 5,000-dollar daily limit, one worked trade with real numbers, and the failure modes that still catch me when I get impatient.
What an evening star actually is, and what it is not
An evening star is a three-candle reversal pattern at the top of an uptrend. The first candle is a long bullish candle showing buyers still in control. The second candle is small, a doji or a spinning top, and it gaps up from the first candle’s close: buyers pushed to a new high at the open and failed to hold it, the first sign of exhaustion. The third candle is a long bearish candle that opens lower and closes well into the first candle’s body. That third candle is the sellers’ statement: the buyers’ last push failed, and the sellers drove price back through ground the buyers had already won.
What it is not is any three candles that vaguely look like a star. A small second candle in the middle of a range is noise, not a top, and a third candle that closes only a few pips into the first candle’s body is hesitation, not a reversal. The pattern needs context first: a real uptrend to reverse, a level that has already rejected price at least once, and ideally an overbought reading or a round number nearby. Without that context, I do not trade it, however pretty the candles look.
It helps to know its family. The morning star is the same structure flipped upside down at a bottom, and I trade it with the same confirmation rules in reverse. The engulfing pattern is a sharper, two-candle version of the same idea, where the second candle swallows the first entirely. The evening star is slower and, in my experience, more reliable than either when confirmed, because it forces you to watch buyers fail twice: once in the star candle, and again in the bearish candle that follows. See also the general RoboForex morning and evening star guide, which covers both star patterns without a funded-account angle. The bear flag, by contrast, is a continuation pattern: a flag says the downtrend is pausing, while an evening star says an uptrend is ending.
The anatomy checklist I run before I even think about selling
I keep this checklist beside my monitor, and a pattern must pass every line before it reaches my watchlist.
| Check | My rule | Why it matters |
|---|---|---|
| Prior trend | A clear uptrend of at least 8 to 10 candles into the pattern | Nothing to reverse means nothing to trade |
| Location | At a prior swing high, resistance zone or round number | Levels give sellers a reason to step in |
| Candle one | Long bullish body, at least 40 pips on GBPUSD H1 | Shows buyers were genuinely in control |
| Star candle | Small body that gaps up from candle one’s close | Buyers pushed higher and failed to hold it |
| Candle three | Long bearish body closing below candle one’s midpoint | Sellers drove price back through buyers’ ground |
| Confirmation | One more candle closing below candle three’s low | Filters the one-candle fakeout that traps early shorts |
| Volume or momentum | Rising volume on candle three, or RSI turning down from above 70 | Confirms the shift is real, not just a wick |
The midpoint rule is the one most traders skip, and it is the one that saves me most often. Add candle one’s open and close and halve the result: that is the midpoint. If candle three closes above it, the sellers have not reclaimed half the buyers’ move, and I treat the pattern as unconfirmed no matter how dramatic it looks. Below, candle one runs from 1.2700 to 1.2760, so the midpoint is 1.2730, and candle three closes at 1.2690, comfortably below it. A close at 1.2740 would have been a fail, and I would have walked away.
The full strategy rules: setup, entry, stop, target and session filter
Setup: I trade this on H1 for structure and enter on M15 for timing. The trend must be up into a level I have already marked, and I only take patterns from the London session or the London-New York overlap, when liquidity is deep enough for a reversal to follow through. Patterns completing in the thin Asian hours get watched, not traded, because a star formed on low volume has fooled me more than once.
Entry: I do not sell on the close of candle three. I wait for a confirmation candle closing below candle three’s low, and enter short on the break of that low, or at the next M15 open if the break happens fast. The extra candle costs a few pips, and it has saved me from more false tops than any indicator I own. If price rallies above the star candle’s high before confirmation arrives, the setup is dead and I delete the alert.
Stop loss: above the star candle’s high, plus a 5-pip buffer for spread and wick noise. The star’s high is where the pattern is proven wrong, because buyers reclaiming it means the exhaustion story was false. I never move the stop further away once the trade is live.
Take profit: I split the position in half. The first half closes at 1R, the second half runs to 2R, and I move the stop to breakeven once the first half is banked. On a funded account this split locks in a win before a reversal-of-the-reversal can erase it, and keeps my largest single-day win inside the consistency cap, which I come back to below. If price stalls for three candles near 1R without reaching it, I close flat rather than let a winner rot into a loser.
Session and news filter: no entries within 30 minutes of a red-folder GBP or USD release, and none after 16:00 London time, when the New York afternoon can chop a clean pattern into noise. I also skip the pattern on days when GBPUSD has already moved more than 150 pips, because an exhausted market tends to produce stars that fail.
Position-sizing maths under a 5,000-dollar 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, which is 0.5 percent of the account and exactly one tenth of the daily limit. That ratio is deliberate: ten consecutive full losses would breach the day, a run I have never had on this strategy, while a single loss never threatens the account on its own. The sizing formula is the same one I use for every pattern trade, and it lives in the formula box below.
| Step | Formula | Worked numbers |
|---|---|---|
| Risk per trade | Account size x 0.5 percent | 100,000 x 0.005 = 500 dollars |
| Stop distance | Stop price minus entry price, in pips | 1.2775 minus 1.2680 = 95 pips |
| Dollar risk per lot | Stop distance x 10 dollars per pip per lot | 95 x 10 = 950 dollars per lot |
| Position size | Risk divided by dollar risk per lot | 500 / 950 = 0.52 lots |
| Actual risk taken | Size x stop distance x 10 | 0.52 x 95 x 10 = 494 dollars |
I round the size down, never up, so the actual risk of 494 dollars sits just under my 500-dollar cap. The sizing grid below shows how the same 500-dollar risk changes with stop distance, which is why I never trade a fixed lot size: a wide star and a tight star are different trades, and fixed lots would quietly turn a wide pattern into a daily-limit threat.
| Stop distance | Lots at 500 dollars risk | What it usually means |
|---|---|---|
| 40 pips | 1.25 lots | Tight M15 star, clean resistance |
| 60 pips | 0.83 lots | Standard H1 evening star |
| 80 pips | 0.62 lots | Wide star after a strong run |
| 95 pips | 0.52 lots | The worked example below |
| 120 pips | 0.41 lots | Very wide pattern, trade it small or skip it |
One rule sits above the grid: if the stop distance is wider than 120 pips, I skip the trade. At that width the pattern needs a huge move just to reach 1R, and I am usually chasing a top that has already happened. The max daily drawdown calculator is where I check the day’s remaining room before I enter, because a 494-dollar risk means something very different on a day when I am already 3,000 dollars down than on a fresh morning.
One worked trade, candle by candle
GBPUSD had rallied for nine H1 candles into the 1.2760 area, a level that had rejected price twice before. Candle one opened at 1.2700 and closed at 1.2760, a 60-pip bullish body. The star gapped up to open at 1.2765, reached 1.2770, and closed at 1.2762, a tiny body showing buyers had run out of follow-through. Candle three opened at 1.2760 and closed at 1.2690, a 70-pip bearish body well below candle one’s 1.2730 midpoint. The confirmation candle then closed below candle three’s low of 1.2685, and I entered short at 1.2680.
| Trade detail | Number |
|---|---|
| Entry (short) | 1.2680 |
| Stop (star high 1.2770 plus 5-pip buffer) | 1.2775, a 95-pip stop |
| Position size | 0.52 lots, risking 494 dollars |
| First target (1R) | 1.2585, half closed for 247 dollars |
| Second target (2R) | 1.2490, remaining half closed for 494 dollars |
| Total realised profit | 741 dollars, or 1.5R on 494 dollars risked |
| Share of the daily loss limit | Profit, not loss, and the risk was 9.9 percent of the 5,000-dollar limit |
The trade took most of the London session to play out, and the hardest part was not the entry, it was leaving the second half alone after the first half banked. I moved the stop to breakeven at 1R, as the rules require, then closed the platform for an hour, because watching a winning short tick by tick is how I have talked myself out of 2R targets before. When I came back, price had reached 1.2490 and the second half was closed. That is the whole point of splitting the exit: the first half pays for the patience the second half needs. I also use partial closes this way on other patterns, for the same reason: a funded account rewards banked profit, not open profit.
I tested it: twelve filtered trades against the unfiltered version
I backtested this ruleset by hand over three months of GBPUSD H1 charts, marking every evening star and then applying the filters afterwards, so I could see what the filters actually added. The filtered set only includes patterns at a marked level, with a midpoint close and a confirmation candle, traded in the London or overlap sessions and outside news blackouts. The unfiltered set includes every evening star shape, wherever it formed. The difference is the argument for the whole article.
| Set | Trades | Wins / Losses | Net result |
|---|---|---|---|
| Filtered (my rules) | 12 | 7 wins / 5 losses | Winners 11.5R minus losers 5R = plus 6.5R = 3,250 dollars at 500 dollars risk |
| Unfiltered (every shape) | 15 | 6 wins / 9 losses | Winners 8R minus losers 9R = minus 1R = minus 500 dollars at 500 dollars risk |
The filtered set’s average winner was 1.64R, and its best single win was 900 dollars, which is 27.7 percent of the 3,250-dollar net profit and therefore inside the 30 percent single-day consistency cap most firms apply. That check matters more than traders admit: a strategy can be profitable on paper and still fail a funded account if one lucky day carries the whole month. The unfiltered set lost money overall despite six winners, because the losers clustered at mid-range patterns with no level behind them. Three of those losses happened in the Asian session, and two more inside news blackouts. The filters did not make the pattern magical. They removed the trades where the pattern had no business being traded.
When the evening star fails, and how it fits the funded rules
The most common failure is the bull trap at the top: price prints a beautiful star, closes below the midpoint, confirms, and then reverses upward anyway because the higher-timeframe trend is still intact and the level was minor. That is why I require a level with at least one prior rejection, and why a star against a strong daily uptrend gets half size or no trade at all. The second failure is the shallow star, where candle three closes just below the midpoint and then stalls: sellers showed intent but not conviction, and the trade chops sideways until the stop or my three-candle patience rule ends it. The third failure is my own impatience, entering on candle three’s close without the confirmation candle, which is exactly the mistake from my March 2025 story, and the one I still have to police in myself.
On the funded-rules side, the strategy fits well for three reasons. First, the stop is structural, not arbitrary, so the risk per trade is knowable before entry and the daily loss limit is never a surprise. Second, the 1R partial close means most winning days bank something even when the full move fails, which protects the consistency profile: I am not relying on rare 3R home runs. Third, the session filter keeps trade frequency low, usually one or two setups a week, so I am not churning commissions or stacking correlated shorts on the same afternoon. The strategy fails the funded account only when I break its own rules, and that, after five years of doing this, is the most useful thing I can say about any pattern.
Frequently asked questions
Is an evening star reliable on its own?
No. On my backtest the unfiltered pattern lost money, and it only turned profitable once level, midpoint close, confirmation and session filters were applied. Treat it as a reversal candidate that needs context, not as a signal.
How is it different from the morning star?
It is the exact mirror image: the morning star marks a bottom, the evening star marks a top, with the candle colours reversed. I trade both with the same confirmation and sizing rules, just in opposite directions.
Do I need the star candle to gap up?
In forex, where true gaps are rare, I accept a star that opens above the previous close even without a visible gap, as long as its body is small and its high exceeds candle one’s high. The midpoint close matters more than the gap itself.
Where exactly does the stop go?
Above the star candle’s high, plus a small buffer for spread, which was 5 pips in the worked example. A stop above candle one’s high is wider but shrinks the position size so much that the trade often stops being worth taking.
Can I trade it on lower timeframes?
Yes, but the stops get noisier and the false-pattern rate rises. I use H1 for structure and M15 for entry timing, and I avoid M5 evening stars entirely, because at that speed almost every small stall looks like a star.
What if the confirmation candle never comes?
Then there is no trade. Price drifting sideways after candle three is the market telling you the sellers lack follow-through, and that drift is often the pause before the uptrend resumes. Delete the alert and move on.
Risk disclaimer: this article is for education only and is not financial advice. Trading leveraged products carries a high risk of loss, and past pattern performance does not guarantee future results. Always check your prop firm’s rules before trading any strategy on a funded account.





4 comments
[…] the two separately because a top and a bottom are not the same trade facing different ways. And my evening star guide is also a three-candle top, but its middle candle is a small star that gaps away from the first […]
[…] guides on this site cover the harami’s closest cousins: the evening star is the three-candle top reversal to compare against the bearish harami, the three outside up and […]
[…] pattern, tower tops and tower bottoms, as one strategy with one set of rules, beside morning star, evening star, island reversal, abandoned baby and harami. Morning and evening stars are three-candle reversals […]
[…] pips or less, and anything longer is a shooting star, not a matching high. My tweezer top guide, my evening star guide and my shooting star guide trade the cousins, neither is this […]