Search for reversal candlestick patterns and you get lists: the top ten, the top fifty, every named shape with a drawing and a sentence. The lists are not wrong. They are answering the wrong question, because a reversal pattern is not a shape. It is a shape in a location. The identical candles, drawn to the pixel, are a reversal in one place, a continuation in another, and nothing at all in a third. The only thing that changes between those three readings is what the chart did before the candles formed.

So this is not another list. It is the post about how to tell which of the three you are looking at. The full per-pattern write-ups already exist in the bullish list and the bearish list; here the patterns get one compact table, and the space goes to the decision those posts assume you have already made.

What separates a reversal pattern from a continuation pattern?

Not the candles. The prior trend.

StockCharts states the rule plainly: for a pattern to qualify as a reversal pattern, there should be a prior trend to reverse, with bullish reversals requiring a preceding downtrend and bearish reversals a prior uptrend. It goes further and gives the case that catches most people out: a bullish engulfing that forms at new highs cannot be a reversal, because there is no decline to reverse. Same two candles, and the reading flips to continuation.

That is the whole mechanism. A reversal pattern records a session, or two or three, in which the side that had been winning stopped winning. The record is only meaningful if a side had been winning. A long lower wick after five days of lower lows says sellers pushed and were absorbed. The same wick after five days of sideways drift says price wandered to the bottom of the range and came back, which is what ranges do.

The proof: two pairs of identical candles

Candlestick vocabulary already contains the proof, because it gives opposite names to identical drawings.

The hammer is a small body near the top of its range with a lower wick at least twice the body’s height. The hanging man is a small body near the top of its range with a lower wick at least twice the body’s height. Bulkowski’s identification guidelines list the same single candle for both; the only line that differs is the prior trend, downward for the hammer and upward for the hanging man.

A hammer drawn from the pattern's real open, high, low and close proportions Hammer
Hammer, one candle, after a decline. Drawn to the real proportions the formation rules describe. Full hammer guide
A hanging man drawn from the pattern's real open, high, low and close proportions Hanging man
Hanging man, one candle, after a rise. Drawn to the real proportions the formation rules describe. Full hanging man guide

Look at the two drawings without their labels and you cannot tell them apart, because there is nothing to tell apart. The hammer reads as a failed sell-off because it sits at the bottom of a decline. The hanging man reads as a warning because the same sell-off happened at the top of an advance, where sellers appearing at all is the news.

The second pair runs the same way. The inverted hammer and the shooting star are both a small body near the bottom of the range under a long upper wick. After a decline the rally-and-give-back is notable because buyers showed up in a market that had none. After an advance the same session is buyers being sold into at the high. Bulkowski’s tested results carry a useful warning here: he labels the behaviour of the hanging man and the shooting star as near random, and finds the inverted hammer more often continues lower than reverses. The shape on its own settles very little, which is the point of everything that follows.

The reversal patterns, by candle count

Every pattern below needs the trend in its “needs” column before it means anything, and every one has a close that ends the read. The names link to the entity pages, and the two roundup posts carry the full mechanics.

PatternCandlesNeeds before itInvalidated by a close
Hammer1a declinebelow the hammer’s low
Hanging man1an advanceabove the hanging man’s high
Inverted hammer1a declinebelow its low
Shooting star1an advanceabove the shooting star’s high
Dragonfly doji1a decline, or a level that has heldbelow the doji’s low
Gravestone doji1an advance, or a level that has failedabove the doji’s high
Bullish engulfing2a declinebelow the engulfing candle’s low
Bearish engulfing2an advanceabove the engulfing candle’s high
Dark cloud cover2an advance, and a real opening gapabove the dark cloud candle’s high
Tweezer bottom2a declinebelow the shared low
Tweezer top2an advanceabove the shared high
Morning star3a declinebelow the middle candle’s low
Evening star3an advanceabove the middle candle’s high
Three white soldiers3a decline or a long basebelow the first candle’s open
Three black crows3an advance or the top of a rangeabove the first crow’s open

Two shapes are missing on purpose. The plain doji and the spinning top record indecision, not a change of control. They become part of a reversal only when a trend precedes them and a decisive candle follows, which is what the morning and evening stars are. The three-candle patterns post looks at that group on its own: what the third candle adds, and the close that ends each one.

Read the table by column, not by row. The “needs” column is the same three words repeated, because the requirement is the same for every pattern. The invalidation column is where the patterns actually differ, and it is the number worth writing down before the next candle opens.

What do the same shapes mean inside a trend?

They describe the trend carrying on, and the clearest example is the candle that has no wicks to argue about.

A marubozu drawn from the pattern's real open, high, low and close proportions Marubozu
Marubozu, one candle, in any trend. Drawn to the real proportions the formation rules describe. Full marubozu guide

A marubozu opens at one extreme and closes at the other. StockCharts describes the white version as buyers controlling the price action from the first trade to the last. Bulkowski’s guidelines list its prior-trend requirement as “none required” and classify it as a continuation of the prevailing trend, while calling its tested behaviour close to random. That combination is honest: the candle records one session of total control by one side, it says that side is still in charge, and it makes no promise about tomorrow.

The subtler continuation is the reversal shape at the wrong end of the move. A hammer that forms after a three-day pullback inside an established advance is not a bottom reversal; there is no decline of any standing to reverse. What it records is the pullback being bought, which is the advance continuing. The engulfing-at-new-highs case StockCharts raises is the same thing. Same candles, and the location has rewritten the sentence.

What does “no pattern” look like?

A range. When price has been moving sideways between a ceiling and a floor, the shapes in the table still appear, constantly, and they describe nothing about direction. A hammer at the bottom of a range says the floor held again. A shooting star at the top says the ceiling held again. Neither is a reversal, because there was no trend, and neither is a continuation, because there is nothing to continue. They are the range being a range.

That is not useless information. It tells you where the edges are, and drawing those levels properly is its own skill. It is just not what a reversal pattern claims to be, and calling it one is how a chart ends up covered in labels that all cancel out.

How do you know there was a trend to reverse?

This is the check almost everyone skips, and it has to be done before the pattern is named, not after. The minimum is a sequence, not a day. For a bullish reversal to be on the table, the decline into the pattern should show at least two successive lower highs and two successive lower lows; for a bearish reversal, the mirror, higher highs and higher lows running into the pattern. That is the ordinary definition of a trend, the one Wikipedia gives for trendlines and the one StockCharts lists as “each reaction peak and trough is lower than the previous”. One red candle is not a downtrend. A gap down on a headline is not a downtrend. A single down day followed by a hammer is two days of noise wearing a pattern’s name.

The trend does not have to be large. StockCharts is explicit that it need not be a major move, only that price has been going one way for the short term or at least over the last few days, and it offers three workable tests: price on the far side of its 20-day exponential moving average, the sequence of peaks and troughs, or price on the far side of a trend line. On the trend line, its own rule applies: two points draw it and a third confirms it, so a line touched once is a guess. Pick one test and apply it the same way every time. What matters is that the test is done before the candle is labelled, because the reading you get from the shape alone is, on Bulkowski’s numbers, not far from a coin toss, and the prior trend is the only thing that makes it more than that.

The step that comes after the location

Once the trend is established and the shape has formed in the right place, the pattern is still a claim about one session, not a fact about the next. StockCharts puts the window for confirmation at one to three days after the pattern, on the grounds that candlestick patterns are short-term and usually effective for one to two weeks. The confirming candle, what it looks like for each pattern, and what waiting for it costs are covered in the confirmation candle post, and it is the natural next read.

The order, then, is fixed: establish the trend, identify the shape, write down the invalidation close, wait for confirmation. Skip the first step and the other three are being applied to a drawing.

What this post is not

It is not a ranking of reversal patterns by reliability, and none of the sources used here support one without conditions attached. Published figures shift with the market, the timeframe and the exact definition tested, and the honest summary from the largest public tally is that most single candles behave close to randomly on their own. That is not a reason to ignore them. It is the reason the location check exists.

The full set of drawings, including the indecision patterns this post leaves out, is in the candlestick pattern library. If reading the four prices in a single candle is still new, start with how to read a stock chart.

Sources

Every figure above was checked against these before publishing.