Morning Star Pattern: The Exact Rule and Why It Almost Never Fires
A morning star is three bars telling a story of exhaustion. Sellers post a long red session, the market then goes quiet in a tiny bar that trades entirely below where those sellers finished, and buyers arrive with a long green bar that closes back through the middle of the original red body.
Every guide draws that picture with a comfortable gap under the middle bar. Real tape is stingier. Running Vike's batch_morning_star over 479,749 daily candles from 376 Binance pairs returned six detections, and the qualifying gaps measured between four and thirteen basis points. Below is the exact rule, what makes it this rare, and six real instances found on four-hour candles, one of which lost.
What a morning star is actually saying
Three sessions, and the middle one is the interesting one. Selling was in control, and then the market simply stopped trading down there — a whole session spent in a narrow range beneath the previous close, with neither side able to move it.
Then demand shows up and reclaims half the damage in a single bar. Read as order flow, that sequence is supply exhausting itself in silence and being replaced, rather than being overwhelmed in a fight. Note which bar is the pivot: the quiet one, not the green one.
Compare it with a bullish engulfing, where one green body swallows one red body and the handover is immediate. A star spends a whole session in limbo first, which is why traders trust it more — and why it turns up so much less often.
How to identify a morning star: the exact rules
batch_morning_star measures. The shaded band is the first bar’s body; the lower dashed line is the level the star’s high must stay under, and the upper one is the body midpoint the third close must clear. The star is grey because its colour is never read.Six conditions, checked in strict order, and only three of the four price columns are ever consulted.
- Bar one is red, and its body is at least the size of the 10-bar average body.
- Bar two is tiny.
- Its body comes in under 0.3 times that same average, and nothing in the crate cares whether the little bar is red, green or perfectly flat.
- Its high must sit under where bar one closed — the entire range, not merely the body.
- Bar three is green, with a body of average size or better.
- That third close has to clear the midpoint between bar one's open and its own close.
How Vike computes it.batch_morning_starreads three bars at a time and stamps the last one+100when all of that holds. The average body it measures against is a plain 10-bar mean which already includes these three candles, so an unusually violent red bar raises the size that both outer bars have to clear.
Notice where that midpoint comes from. Averaging an open against a close gives you the centre of a body, never the centre of a range, so however long the shadows on bar one happen to be, they cannot shift the level bar three must beat.
The clause that makes it rare
One condition does essentially all the work, and it is the one about the middle bar's high.
Across 1,495,182 four-hour candles I counted 66,020 cases where a long red bar was followed by a tiny-bodied one. Exactly 187 of those kept the small bar's whole range beneath that closing price — 0.28%, meaning the clause discarded 99.7% of everything reaching it. The remaining conditions then took 187 down to 25.
Why so brutal? A market that never closes opens the next bar at the previous close about 37% of the time and a tick either side of it the rest of the time. This rule wants a gap, and it wants that gap never to be filled for an entire session. Crypto obliges rarely and barely: the median qualifying gap in my scan was 0.079%, and the widest was 0.216%.
So the diagram in every textbook, with daylight under the star, is drawn from equity charts that shut overnight. Here it is satisfied by a rounding error, or not at all.
Why this page uses four-hour candles
Daily crypto candles cannot really produce this pattern, and I would rather show you the measurement than quietly change timeframe.
Over those 479,749 daily bars from 376 pairs, batch_morning_star fired six times: ZEC, API3, GAS, SENT, FIDA and DCR. Its mirror, batch_evening_star, fired zero times on the same tape. Six examples with no failure among them would make a dishonest page.
Shorter bars leave a gap less time to fill, so the same untouched rule becomes survivable: 25 detections in 1.5 million four-hour candles, roughly one bar in sixty thousand. Every chart on this page comes from that scan. The rule was never loosened, and running the fetch script with --daily reproduces the six-instance daily result exactly.
What the rule does not say
Star colour is never tested. Red, green or a perfect doji all satisfy the middle clause, because only the body's size and its position are measured.
No gap is required on the third bar either, which contradicts most illustrations. Buyers may open exactly where the star closed and still qualify, provided the close clears that midpoint.
And nothing anywhere in the function knows what came before. There is no downtrend requirement, no support level, no volume test — a morning star printing mid-range is the same +100 as one printing at the low of the year. Trend context is your job, not the detector's.
One more asymmetry worth carrying away: the function loads the open, high and close columns and never touches the low. Whatever wick hangs beneath the star is invisible to it.
Morning star vs evening star
Vike's crate writes these as two separate functions rather than one function with a sign, which is how the engulfing pair works. Read side by side, they mirror perfectly: red becomes green, the gap clause flips direction, and "closes above the midpoint" becomes "closes below" it.
That mirror even swaps the price column. batch_morning_star destructures open, high and close; batch_evening_star destructures open, low and close. Each is blind to the shadow on the opposite side of the star.
Market behaviour refuses to mirror, though. On identical daily tape, 30 candidates cleared the morning star's gap clause against 7 for the evening star, and on four-hour bars it was 187 against 136. Crypto gaps downward far more readily than it gaps up, so the bullish version of this pattern is the more findable one despite both rules being equally strict.
How to trade a morning star
Trade the level it marks, and be honest about what the pattern costs to confirm.
The textbook trigger barely exists here. Waiting for the next bar to close above the pattern high sounds prudent, and across my 25 four-hour detections it happened exactly twice. Demand it and you pass on 92% of them, so either take the third close or accept that this setup is mostly unavailable.
Stop under the star. The lowest print of the three bars is the invalidation, and because the star is a quiet bar, that stop tends to sit close — the one genuine structural advantage this shape has over a two-bar reversal.
Grade the first bar. When bar one is enormous relative to the average, the midpoint waiting to be reclaimed sits a long way off, which makes the signal stronger and much later. I've found the tighter versions easier to manage.
Ignore it outside a decline. No trend test exists in the code, so the entire burden falls on you. I skip any instance that prints inside a range, and that single filter throws away most of what the scanner shows me.
Where morning stars fail
TUT on 2026-07-16 is the failure shown here, and I picked it precisely because nothing about it was marginal. A red first bar 2.68 times the average body, a star at 0.19, a gap of 0.079%, and a green bar clearing the midpoint by 1.35% — comfortable numbers on every clause. Price then dropped a further 2.6%.
An ugly, borderline detection failing would teach you nothing. This one had the largest first body of anything on the page and still described only the three bars it measured. Of the 22 detections old enough to grade, 12 turned and 10 did not, which is about what a shape carrying no trend test deserves.
That is the honest summary of this pattern on crypto tape. It is real, the engine finds it, and the version the tape produces is a much thinner thing than the version in the diagrams. When I trade one, the level underneath it is doing more work than the candles are.
Frequently asked questions
Does the middle candle have to be a doji?
No. Any body under 0.3 times the 10-bar average qualifies, in any colour. There is a separate detector, batch_morning_doji_star, that tightens the threshold to 0.1 times the average and changes nothing else — so every morning doji star is also a morning star, but not the other way round.
Is a gap really required? Yes, and strictly. Bar two's high has to stay beneath bar one's closing price, which is a full-range gap rather than a gap between bodies. That single clause eliminated 99.7% of otherwise-qualifying setups in my scan.
Why does the third bar only have to reach the midpoint? Because that is where the crate draws the line: the centre of bar one's body. Reaching it means buyers took back more than half of the selling session, which the rule treats as sufficient without demanding a full recovery.
How reliable is the morning star? Too rare on this data to make a confident claim, which is itself the finding. Twelve of the 22 gradeable four-hour detections turned and ten did not, on a total sample of 25 across 376 pairs — nothing anyone should build a strategy on.
What is the bearish version? The evening star, written as a genuine mirror. If you want a bullish three-bar pattern that actually fires often enough to study, three white soldiers is the better subject.
This is educational material, not financial advice. Candlestick patterns describe probabilities, not certainties, and every real trade carries real risk — size positions so a wrong read costs little.