Candlestick patterns and what they describe

A name is a rule, not an extra fact 

Each of these formations is a comparison between prices already drawn. The open, high, low and close of one period are compared with each other, or with the same four values from the periods beside it, and if the comparison comes out a certain way the arrangement has a name. Nothing is measured that was not already on the chart. The name is a compression device, a way for two people to refer to one arrangement in a word rather than five sentences, not a second reading of the market taken alongside the first.

The names are inherited. Most were in use before electronic charting, on markets that shut overnight and reopened the following morning, and they were carried into currency and index charts unchanged. That history matters later, because a few of the definitions quietly assume a market that stops.

One consequence lands immediately. A formation inherits every limitation of the candles it is built from, and the lesson on candlestick anatomy established those: the order in which prices happened inside a period is gone, the open and the close exist because a session clock cut the period there, and how much trading each candle summarises is set by the timeframe. A named arrangement is a summary of summaries, and can be no more precise than the layer beneath it.

Formations built from one candle 

The smallest family is defined by two proportions: how large the body is compared with the range, and where in that range the body sits. Each name below is one of those two proportions taken to an extreme.

• Doji. The open and the close fall at or near the same price, so the body is close to a horizontal line and the drawing is dominated by whatever wicks the period produced.
• Spinning top. A small body near the middle of the range, with wicks of comparable length above and below it.
• Marubozu. A body spanning almost the whole range, so there is little or no wick at either end.
• Hammer and hanging man. A small body near the top of the range with a long lower wick beneath it. The two names describe an identical construction, distinguished only by what preceded it.
• Inverted hammer and shooting star. The mirror image, a small body near the bottom of the range with a long upper wick, again with two names for one shape.
• Pin bar. A later and broader term for any single candle whose wick in one direction dominates its range, with the body pushed to the opposite end.

The pair of names for one shape is worth pausing on, because the doubling is not an accident of vocabulary. The same drawing is called a hammer after a decline and a hanging man after an advance. Construction alone does not decide which word applies, so the naming system already concedes that the surrounding chart does part of the work.

The other thing to notice is how much weight the vague words carry. At or near the same price. A small body. A long wick. No tradition fixes any of them, and a program cannot run on them until somebody chooses a number. Different authors and different screening tools choose differently, and the choice decides whether a period is an instance of the formation or an ordinary candle.

The threshold is not housekeeping. Any published description beginning with the words after a doji appears is only as precise as the threshold that assembled its examples, and that threshold is very often not printed beside it. Two people can read the same sentence, apply it to the same chart, and disagree about which periods it refers to.

Formations built from two candles 

Add a second candle and the comparison stops being internal. Each of these states how the second period's numbers sit against the first period's, in one of three shapes: one candle covers the other, one is contained by the other, or the two share an extreme.

• Engulfing. The second body covers the first body completely and the two candles close on opposite sides of their opens, so they draw in opposite colours. The direction of the second gives the formation its bullish or bearish prefix.
• Harami. The containment reversed, a large body followed by a much smaller one that sits entirely inside it.
• Piercing line and dark cloud cover. The second period opens beyond the first period's close, then closes back past the midpoint of the first body without covering it. Partial coverage rather than full.
• Tweezers. Two adjacent candles reaching almost exactly the same high, or almost exactly the same low, whatever their bodies do.

How coverage is measured is genuinely unsettled. One convention tests bodies only, since the body is where the period opened and closed and the wicks are excursions it did not hold. Another tests the full range, wick tip to wick tip, since the second period has not covered the first unless it traded through everything the first traded through. Both are in print, both are coherent, and they do not agree on the same chart.

Formations built from three candles 

A third candle adds a sequence. The morning star and the evening star are the best known: a large body, then a small body sitting clear of it, then a large body in the opposite direction to the first, closing well back into it. Three white soldiers and three black crows are three consecutive candles in the same direction, each with a substantial body and each closing beyond the previous close. Three inside up and three inside down take a harami and add a third period closing beyond the first body.

Every added candle adds conditions, and every condition adds a threshold. Sitting clear of it, well back into it, a substantial body, closing beyond: each phrase has to become a number before anything can be counted. Strictly defined three candle formations are therefore uncommon on any one instrument, which is why loosened variants circulate. A definition relaxed until instances appear at a comfortable rate has been fitted to the chart rather than tested against it, and the relaxation is rarely stated.

Why the same formation is not the same event twice 

Candlestick traditions conventionally hold that where a formation appears matters more than the formation itself. The same construction is described differently after an extended advance, inside a range that has held for weeks, at a level that has already turned price more than once, or in an unremarkable stretch with nothing around it. That is the earlier material in this module doing its work: the sequence of swing highs and swing lows that makes a trend or a range, and the levels repeated turns leave behind.

Something real sits underneath the convention. A long lower wick at the bottom of a range that has held several times is the same drawing as one in a quiet stretch, but the two sit inside different populations of prior prices. The drawing has not changed. The facts available to describe it have.

The convention also comes at a cost, and a serious one. The level supplying the context was drawn by a person, with the discretion the lesson on support and resistance set out, so a formation at a level is at a level because somebody put the line there. Adding a context requirement makes a definition harder to falsify rather than more precise, since any instance followed by nothing can be excluded afterwards on the ground that the context was wrong. Practitioners disagree about this. One camp treats context as the condition that makes the vocabulary usable at all, another as an escape clause protecting it from ever being wrong, and both readings fit the same literature.

The conditions under which the description fails 

Every failure below is structural. None is a matter of applying the definitions badly, and all of them are present even when the definitions are applied perfectly.

• Nothing is a formation until the last candle closes. A running period has a provisional body, because its close has not happened, so an arrangement that satisfies a definition partway through can fail it at the boundary.
• The session clock creates and destroys them. Bodies are made of opens and closes, which exist because a boundary was drawn. Moving that boundary, which platforms and data vendors do differently, changes every body on the chart, so a body based formation present on one clock is absent on another with no trade having differed.
• The timeframe decides what exists. A three candle formation is part of one candle on a higher timeframe and a long sequence on a lower one. No timeframe is the correct one, so presence is always a statement about a chosen resolution.
• Some definitions assume a market that stops. Piercing line, dark cloud cover and others are written around a period opening beyond the previous close, which requires the kind of gap an overnight break produces. Currency markets run continuously through the working week, so most periods open at or within a fraction of the previous close, and those conditions are either almost never met or met so marginally that the formation is trivial.
• Wick based shapes are feed dependent. A brief extreme reaching one venue's book may not appear in another provider's aggregated stream, so hammers, pin bars and tweezers can be present on one chart and absent on an honest chart of the same instrument.
• Presence is a function of how much is being looked at. Dozens of named formations, across hundreds of instruments and several timeframes, guarantee that some formation is present somewhere at every moment. Finding one is evidence about the size of the search, not about the instrument.

Figures ranking these formations are not reproduced here 

Pattern references and vendor material frequently publish tables scoring each named formation by how often it was followed by a movement of some size in some direction. Nothing of that kind appears on this page, and the omission is deliberate rather than an oversight to be filled in later.

The first reason is methodological. A score of that kind is not a property of the formation. It is the joint product of choices rarely printed beside the number: the threshold deciding which candles counted as instances, the instruments and span of history sampled, the timeframe, what counted as a movement, how long after the formation the measurement was taken, and what ended it. Change any one and the number changes, so a figure produced under one set of choices does not transfer to a chart drawn under another.

The second reason is simpler. This page has no verified source for such a figure and no way to attach an as at date to one, and an unsourced statistic is a fabricated figure whether or not it concerns a broker. The claim is absent rather than softened. Where a ranked table is met elsewhere, the questions above are what decide whether its numbers mean anything for a different chart, and they are answerable only from the methodology behind it.

Where practitioners disagree 

The deepest disagreement is whether these arrangements carry information at all. One tradition reads them as a compressed trace of activity: a long lower wick closing near the top of the range is a period in which selling reached a level and was met there, and the shape is the residue. Another holds that the boundaries producing the bodies were set by a clock rather than by the market, so a body is partly an artefact of that clock and any regularity among bodies may be a regularity of the partition. Both explain the same chart, which is why the argument has not resolved. How a quoted price is assembled before it is ever drawn as a candle is treated in the guide to price formation.

A quieter disagreement is whether the vocabulary earns its place. Many chart readers hold that the named formations are a coarse way of noticing two things measurable directly: how large a period's range is against recent ranges, and where in that range the period closed. On this reading the names add memorability and subtract precision, since a stated proportion can be checked and a small body cannot. Those who hold it still read candles closely. They simply do not name what they are reading.