Chart & Candlestick Pattern Guide

Every formation the scanners look for, drawn and explained. Each entry says what identifies the shape, what would confirm it, what would invalidate it — and, for the harmonics, the exact tolerance this scanner allows, read straight out of the detector rather than copied from a book.

A pattern describes what price has already drawn. It is geometry, not a forecast — several of the best known names are measured close to a coin flip, which is why the candlestick section below carries the studied reversal rate beside each one instead of leaving the name to speak for itself.

66 patterns

Harmonic 8

GartleyReversal at D

The original five-point harmonic. Price makes an impulse, retraces most of the way back, rallies, then falls to a point that sits just inside where the impulse began.

How to identify it: X→A is the impulse. B retraces close to 0.618 of it. C rallies partway back up the A→B leg. D completes near 0.786 of X→A — inside the impulse, never past X. Drawn out it reads as an M in a downtrend or a W in an uptrend.
What confirms it: D is a zone, not a turn. The pattern earns attention only once price reacts there — a rejection wick, a reversal candle, a shift in momentum. A touch on its own confirms nothing.
What invalidates it: A decisive move through X. At that point D was not the completion of anything; the structure was something else.

What this scanner accepts

B, against the X→A leg	0.550–0.700 (ideal 0.618)
C, against the A→B leg	0.382–0.886 (ideal 0.618)
D, against the B→C leg	1.130–1.618 (ideal 1.272)
D, against the X→A leg	0.740–0.830 (ideal 0.786)

A shape measuring near the ideal scores high; one scraping the edge of a band is reported with a low quality rather than hidden.

ButterflyReversal beyond X

An extension harmonic. Where the Gartley completes inside its own impulse, the Butterfly runs past it — D lands beyond X.

How to identify it: B sits deep, near 0.786 of X→A. C retraces the A→B leg. D is projected past X, typically around 1.272 of X→A. The deep B and the overshoot together are what separate it from a Gartley.
What confirms it: Wait for the whole zone to be tested and then rejected. These complete into an extended move, so the first touch is often not the low or high.
What invalidates it: Price settling beyond the extension cluster rather than reacting to it.

What this scanner accepts

B, against the X→A leg	0.700–0.850 (ideal 0.786)
C, against the A→B leg	0.382–0.886 (ideal 0.618)
D, against the B→C leg	1.618–2.618 (ideal 2.000)
D, against the X→A leg	1.200–1.700 (ideal 1.272)

A shape measuring near the ideal scores high; one scraping the edge of a band is reported with a low quality rather than hidden.

BatReversal at 0.886

A shallow B and a deep D. The Bat is defined by where it finishes — right up against the start of the impulse, without quite reaching it.

How to identify it: B is shallow, around 0.45 of X→A and always under 0.618 — that is the first thing to check, because a deeper B makes it a Gartley instead. D completes at 0.886 of X→A, which is the pattern's signature.
What confirms it: The 0.886 level sits close to X, so the margin between a completed Bat and a broken one is thin. Reaction at the zone matters more here than in the wider harmonics.
What invalidates it: Any close beyond X. There is very little room between D and failure.

What this scanner accepts

B, against the X→A leg	0.360–0.520 (ideal 0.450)
C, against the A→B leg	0.382–0.886 (ideal 0.618)
D, against the B→C leg	1.618–2.900 (ideal 2.000)
D, against the X→A leg	0.845–0.920 (ideal 0.886)

A shape measuring near the ideal scores high; one scraping the edge of a band is reported with a low quality rather than hidden.

CrabDeep extension

The most extended of the family. D projects far past X, which makes the completion zone a long way from the rest of the structure.

How to identify it: B is shallow, 0.382–0.618 of X→A. The C→D leg is the giveaway: it extends 2.24 or more of B→C, carrying D to roughly 1.618 of X→A.
What confirms it: The long final leg means price arrives at the zone with momentum behind it. Evidence that the momentum has stopped matters more than the level being touched.
What invalidates it: Continuation through the zone without a reaction.

What this scanner accepts

B, against the X→A leg	0.382–0.618 (ideal 0.500)
C, against the A→B leg	0.382–0.886 (ideal 0.618)
D, against the B→C leg	2.240–5.000 (ideal 3.618)
D, against the X→A leg	1.550–1.700 (ideal 1.618)

A shape measuring near the ideal scores high; one scraping the edge of a band is reported with a low quality rather than hidden.

Deep CrabDeep B, deep extension

A Crab that begins with a deep retracement. Same extended ending, different opening.

How to identify it: B at 0.886 of X→A — as deep as a Bat ends. D still extends to about 1.618 of X→A. The deep B is the only thing separating it from an ordinary Crab.
What confirms it: As for the Crab: a reaction at the zone, not merely a visit to it.
What invalidates it: Acceptance beyond the extension.

What this scanner accepts

B, against the X→A leg	0.845–0.920 (ideal 0.886)
C, against the A→B leg	0.382–0.886 (ideal 0.618)
D, against the B→C leg	2.240–4.200 (ideal 3.140)
D, against the X→A leg	1.550–1.700 (ideal 1.618)

A shape measuring near the ideal scores high; one scraping the edge of a band is reported with a low quality rather than hidden.

SharkTerminal point past the first leg

A later addition to the family, and measured differently: its completion is read against the first leg rather than against X→A the way the others are.

How to identify it: The C→D leg extends beyond the first leg, finishing between 0.886 and 1.13 of it. Because the terminal point is defined this way, the Shark is scored on different legs from the Gartley family — the scanner checks X→D against X→A.
What confirms it: It completes in a zone rather than at a line. Treat the band as the level.
What invalidates it: Travelling well past 1.13 of the first leg without reacting.

What this scanner accepts

C, against the A→B leg	1.130–1.618 (ideal 1.300)
D, against the B→C leg	1.618–2.240 (ideal 1.902)
D, measured from X against X→A	0.886–1.130 (ideal 1.000)

A shape measuring near the ideal scores high; one scraping the edge of a band is reported with a low quality rather than hidden.

CypherC beyond A, D measured from X→C

The odd one of the family. Its C point pushes past A — which none of the others do — and its completion is measured against the whole X→C swing rather than against X→A.

How to identify it: B retraces 0.382–0.618 of X→A. C then extends PAST A, to 1.272–1.414 of X→A. D retraces 0.786 of the full X→C swing, which lands it between B and C rather than beyond B.
What confirms it: Because D sits inside the structure rather than past it, the usual "has it broken out" read does not apply. Look for the reaction at the level itself.
What invalidates it: A close back through B.

What this scanner accepts

B, against the X→A leg	0.382–0.618 (ideal 0.500)
C, against the X→A leg	1.272–1.414 (ideal 1.330)
D, retraced back from C against X→C	0.740–0.830 (ideal 0.786)

A shape measuring near the ideal scores high; one scraping the edge of a band is reported with a low quality rather than hidden.

ABCDFour points, two legs

The simplest member: two legs of similar length separated by a retracement. Every five-point harmonic contains one, which is both why it appears so often and why it says the least.

How to identify it: A→B is the first leg. C retraces part of it. C→D runs roughly the same distance as A→B. That is the whole pattern.
What confirms it: With only three measurements it is easy to satisfy by accident, so it carries weight only with something else — a level, a trendline, a larger structure it sits inside.
What invalidates it: C→D running far past A→B, which makes it an extension rather than an AB=CD.

Reversal 10

Double TopTwo peaks, one neckline

Price reaches a level, falls away, returns to the same level and fails again. The low between the two peaks is the line that decides it.

How to identify it: Two highs within a few percent of each other, separated by a trough deep enough that they read as two attempts rather than one. The further apart in time, the more the level has been tested.
What confirms it: A close below the trough between the peaks. Before that it is two highs and nothing more — plenty of charts make two highs and keep going.
What invalidates it: A close above the higher peak.

Double BottomTwo troughs, one neckline

The mirror image. Price finds a level twice and the rally between the two lows is the line that decides it.

How to identify it: Two lows at a similar price with a clear rally between them.
What confirms it: A close above the peak between the two lows.
What invalidates it: A close below the lower of the two lows.

Triple TopThree attempts, one level

Three failures at the same level. More testing than a double top, and a more clearly defined line underneath.

How to identify it: Three highs at a similar price with two troughs between them. The lower of those troughs is the line.
What confirms it: A close below the lower trough.
What invalidates it: A close above the highest of the three peaks.

Triple BottomThree attempts, one level

Three failures to break lower at the same level.

How to identify it: Three lows at a similar price with two rallies between them.
What confirms it: A close above the higher of the two intervening peaks.
What invalidates it: A close below the lowest of the three.

Head & ShouldersThree peaks, middle highest

Three peaks where the middle one is the tallest, joined underneath by a line through the two troughs. The best known reversal shape in the literature.

How to identify it: Left shoulder, a higher head, then a right shoulder at roughly the left's height. The neckline runs through the two troughs and need not be flat.
What confirms it: A close below the neckline. The shape without the break is three peaks.
What invalidates it: A close above the head.

Inverse Head & ShouldersThree troughs, middle lowest

The same shape upside down: three lows with the middle one deepest.

How to identify it: Left shoulder, a deeper head, a right shoulder near the left's depth, with the neckline across the two intervening peaks.
What confirms it: A close above the neckline.
What invalidates it: A close below the head.

Three Rising ValleysEach low higher than the last

Three troughs, each one higher than the one before. A description of pressure easing rather than of a level holding.

How to identify it: Three distinct lows in ascending order, with rallies between them.
What confirms it: A close above the highest peak in the sequence.
What invalidates it: A low below the second trough, which breaks the rising sequence.

Three Falling PeaksEach high lower than the last

Three peaks in descending order — the mirror of three rising valleys.

How to identify it: Three distinct highs, each below the previous, with troughs between.
What confirms it: A close below the lowest trough in the sequence.
What invalidates it: A high above the second peak.

Rounding BottomA slow turn, not a spike

A gradual curve from falling, through flat, to rising. There is no single low to point at — that is the point of it.

How to identify it: A long, shallow arc with no sharp turning point. The shape takes time; a quick V is a different thing.
What confirms it: A close above the level where the decline into the arc began.
What invalidates it: A sharp break below the base of the arc.

Rounding TopA slow roll over

The same slow curve the other way up: rising, flattening, then falling.

How to identify it: A long shallow dome with no single sharp high.
What confirms it: A close below the level where the advance into the dome began.
What invalidates it: A decisive push above the dome.

Wedge / Triangle 5

Rising WedgeBoth lines up, converging

Highs and lows both rising, but the lows rising faster, so the range narrows as price climbs. Advance without widening.

How to identify it: At least two touches on each line. Both slope up; the lower is steeper. The narrowing is the pattern — a rising channel has parallel lines and is a different read.
What confirms it: A close below the lower line.
What invalidates it: A close above the upper line, which turns the narrowing into a breakout.

Falling WedgeBoth lines down, converging

Highs and lows both falling, the highs falling faster, so the range narrows on the way down.

How to identify it: Two touches on each line, both sloping down, the upper steeper.
What confirms it: A close above the upper line.
What invalidates it: A close below the lower line.

Symmetrical TriangleConverging from both sides

Lower highs and higher lows at a similar rate. The range closes from both directions and the pattern itself leans neither way.

How to identify it: Falling upper line, rising lower line, roughly matched in steepness, with at least two touches each.
What confirms it: A close outside either line. The shape does not say which — it says the range is closing.
What invalidates it: Price reaching the apex without breaking out: the structure has expired rather than resolved.

Ascending TriangleFlat top, rising lows

Repeated failures at one level while the lows keep rising. Supply sitting at a price, demand arriving earlier each time.

How to identify it: Two or more highs at the same level and two or more rising lows beneath them.
What confirms it: A close above the flat top.
What invalidates it: A close below the rising line, which removes the pressure the pattern describes.

Descending TriangleFlat bottom, falling highs

A level that holds repeatedly while each rally falls short of the last.

How to identify it: Two or more lows at the same level and two or more falling highs above them.
What confirms it: A close below the flat bottom.
What invalidates it: A close above the falling line.

Rectangle 2

Rectangle TopA range after a rise

Price moves sideways between two horizontal lines after an advance. Neither side wins until one line goes.

How to identify it: At least two touches on a flat top and two on a flat bottom, with the move into it having been upward.
What confirms it: A close outside either line.
What invalidates it: The range widening, which makes it something other than a rectangle.

Rectangle BottomA range after a fall

The same horizontal range, entered from above.

How to identify it: Two touches on each of a flat top and a flat bottom, reached by a decline.
What confirms it: A close outside either line.
What invalidates it: The boundaries losing their flatness.

Broadening 2

Broadening TopDiverging lines, after a rise

Higher highs and lower lows at the same time — the range widens instead of closing. Disagreement increasing rather than resolving.

How to identify it: An upper line sloping up and a lower line sloping down, each with at least two touches.
What confirms it: A close outside either line, though in a widening range that level keeps moving.
What invalidates it: The range starting to narrow again.

Broadening BottomDiverging lines, after a fall

The same widening range, arrived at from above.

How to identify it: Diverging boundaries with at least two touches each, entered by a decline.
What confirms it: A close outside either line.
What invalidates it: Convergence returning.

Continuation 4

FlagA pause inside a move

A sharp move, then a short drift against it between two parallel lines. The drift is shallow and brief — a rest, not a reversal.

How to identify it: A steep run in (the pole), then a narrow parallel channel sloping gently the other way. The flag should be short against the pole.
What confirms it: A close out of the channel in the direction of the pole.
What invalidates it: The drift retracing most of the pole, which makes it a reversal rather than a pause.

PennantA converging pause

The same pause as a flag, but the two lines converge instead of running parallel — a very small symmetrical triangle inside a move.

How to identify it: A steep pole, then a brief narrowing between a falling high line and a rising low line.
What confirms it: A close out of the convergence in the direction of the pole.
What invalidates it: The pause growing long enough to become a triangle in its own right.

High & Tight FlagA steep pole, a shallow pause

A flag with an unusually steep pole and an unusually shallow pause. Rare, because both halves have to be extreme at once.

How to identify it: A near-vertical advance, then a pullback that gives back only a small part of it before the range tightens.
What confirms it: A close above the pause.
What invalidates it: A deep retracement of the pole.

Cup with HandleA rounded base, then a small dip

A long rounded recovery back to a previous high, then a short shallow pullback before the attempt on it.

How to identify it: The cup is a smooth U, not a V. The handle is a brief drift lower that stays in the upper part of the cup — a deep handle is a different structure.
What confirms it: A close above the rim, the level both sides of the cup reach.
What invalidates it: The handle retracing into the lower half of the cup.

Momentum 2

Weak ReboundA small bounce inside a fall

A decline, then a rally that recovers only a small part of it. Named for what the bounce failed to do rather than for a shape.

How to identify it: A clear down leg followed by a rally that retraces a modest share of it and stalls.
What confirms it: A move back below the start of the rebound.
What invalidates it: The rebound carrying through most of the decline, at which point it was not weak.

Weak PullbackA small dip inside a rise

An advance, then a dip that gives back only a small part of it.

How to identify it: A clear up leg followed by a shallow pullback that holds.
What confirms it: A move back above the start of the pullback.
What invalidates it: The pullback giving back most of the advance.

Trendline 1

Trendline S/RA line price keeps meeting

Not a formation but a line: one drawn through two or more turning points that price has respected since. It annotates a chart rather than predicting one.

How to identify it: Two touches define a line; the third is what makes it worth drawing. The touches matter more than the slope.
What confirms it: Nothing to confirm — it is a level, not a setup. Its usefulness is in how price behaves when it returns.
What invalidates it: Price closing through it and then staying there.

Candlestick formations 32

The rate is an educational reference label, not a measurement of any one stock or a guarantee. The scanner classifies each formation by its classical price structure and prior trend.

Single

Formation	Direction	Reversal rate	What it is
Hammer	bullish	60%	After a decline: a small body at the top of the range with a clean lower shadow at least twice the body and filling most of it. Reported only on a prominent, heavier-than-usual-volume candle that stands out from the sessions around it — sellers pushed price down through the session and were bought back before the close.
Inverted Hammer	bullish	60%	After a decline: a small body at the bottom of the range with a clean upper shadow at least twice the body and filling most of it. Reported only on a prominent, heavier-than-usual-volume candle — a failed rally inside a downtrend, read as the first attempt to turn.
Hanging Man	bearish	59%	The hammer shape, but after an advance rather than a decline. Same candle, opposite context — which is the whole point: position in the trend is what names it.
Shooting Star	bearish	59%	After an advance: a small body at the bottom of the range with an upper shadow at least twice the body. Buyers drove price up and lost all of it by the close.
Doji	neutral	—	Open and close are effectively the same price. It is indecision, not a direction — which is why it is reported as neutral no matter where it appears.
Dragonfly Doji	bullish	50%	A doji whose open, high and close sit at the top of a long lower shadow. Bulkowski measures it near a coin flip, so it is worth seeing and not worth much on its own.
Gravestone Doji	bearish	50%	A doji whose open, low and close sit at the bottom of a long upper shadow. Also close to a coin flip in testing.
Bullish Marubozu	bullish	—	A long white body with almost no shadow either end: the session opened at its low and closed at its high. A statement of one-sided pressure, not a reversal signal.
Bearish Marubozu	bearish	—	A long black body with almost no shadow: opened at its high, closed at its low.
Spinning Top	neutral	—	A small body with long shadows on both sides. Price travelled a long way in both directions and finished where it started.

Two candle

Formation	Direction	Reversal rate	What it is
Bullish Engulfing	bullish	63%	After a decline: a white body that completely covers the previous black body. Common — Bulkowski ranks it 12th of 103 for frequency — but the move that follows ranks only 84th, so it turns price often and not far.
Bearish Engulfing	bearish	79%	After an advance: a black body that completely covers the previous white body.
Bullish Harami	bullish	53%	After a decline: a small body contained entirely within the previous long body. Barely better than a coin flip in testing — the smaller the second body, the more the literature expects of it.
Bearish Harami	bearish	47%	After an advance: a small bearish real body contained entirely within the previous long bullish body. The name promises a top, but the testing does not: Bulkowski measured it acting as a bullish continuation 53% of the time, so it reverses only about 47% — the one well-known pattern whose behaviour contradicts its name. Treat a bearish read as unconfirmed until price actually breaks the low.
Bullish Harami Cross	bullish	55%	After a decline: a long bearish candle followed by a doji contained inside its real body. It marks indecision after selling pressure and needs bullish follow-through.
Bearish Harami Cross	bearish	43%	After an advance: a long bullish candle followed by a doji contained inside its real body. Like the Bearish Harami, its name over-promises — Bulkowski measured it continuing higher about 57% of the time, so it reverses only around 43%. The doji marks indecision, not a confirmed top; wait for a decisive break of the low.
Piercing Line	bullish	64%	After a decline: price opens below the previous close and closes back above the midpoint of the previous black body. Half the prior session undone in one go.
Dark Cloud Cover	bearish	60%	After an advance: price opens above the previous high and closes back below the midpoint of the previous white body.
Tweezer Bottom	bullish	—	Two sessions that stop at the same low after a decline. The shared level is the whole pattern.
Tweezer Top	bearish	—	Two sessions that stop at the same high after an advance.

Three candle

Formation	Direction	Reversal rate	What it is
Morning Star	bullish	78%	After a decline: a long black candle, a small-bodied candle that gaps down away from it, then a long white candle closing well into the first body. One of the stronger patterns in testing — 6th of 103.
Evening Star	bearish	72%	After an advance: a long white candle, a small body gapping above it, then a long black candle closing at least midway down the first body. Ranks 4th of 103 on the move that follows.
Morning Doji Star	bullish	76%	A morning star whose middle candle is a doji — indecision at the low rather than merely a small body.
Evening Doji Star	bearish	71%	An evening star whose middle candle is a doji.
Three White Soldiers	bullish	82%	Three tall white candles, each opening inside the previous body and closing near its own high. One of the strongest patterns Bulkowski measured — 3rd of 103 by reversal rate — and correspondingly rare, because the rules are strict.
Three Black Crows	bearish	78%	Three tall black candles, each opening inside the previous body and closing near its own low. Ranks 3rd of 103 on the size of the move that follows.
Three Inside Up	bullish	65%	A bullish harami with a confirming third candle that closes above the second. The confirmation is what separates it from the harami alone, which is close to a coin flip.
Three Inside Down	bearish	65%	A bearish harami with a third candle closing below the second.
Three Outside Up	bullish	75%	A bullish engulfing with a third candle closing higher still.
Three Outside Down	bearish	75%	A bearish engulfing with a third candle closing lower still.
Bullish Abandoned Baby	bullish	70%	A doji isolated by gaps on BOTH sides at the bottom of a decline — its whole range below the previous low and below the next low. Very rare, because a true gap on both sides of one session is rare.
Bearish Abandoned Baby	bearish	70%	The same isolation by gaps, at the top of an advance.

Alert

Info