What is a hammer candlestick?

A daily candle compresses an entire session into four numbers: where trading opened, the highest price paid, the lowest price paid, and where trading closed. Most sessions produce a shape that is easy to summarise, because price went one way and stayed there. The hammer is the case where those four numbers preserve a story with two acts, and the shape is legible precisely because the two acts pulled in opposite directions.

In the first act the sellers are winning. Price is marked down from the open, and the move often accelerates as it goes, because resting stop orders below recent lows are triggered and become market sell orders that push price lower still. This is the part of the session that feels, to anyone watching it live, like the beginning of something much worse. In the second act, somewhere well below the open, the selling stops getting paid. Bids meet the offers instead of stepping away from them, the market stops printing new lows, and price is repriced back up through the whole excursion until the close lands near the top of the range.

The long lower shadow is the receipt for that excursion. Trades printed at every price along it, which means real supply came out down there and real money took the other side of it. By the close, everyone who sold in the lower stretch of the range is watching the market trade above their exit. The participants who most wanted out have already gone, at prices below the close, and the ones who bought from them are in front. Demonstrated demand meeting absorbed supply, inside one period, is the entire informational content of the candle.

That also explains why the close has to sit near the high. A session that reclaims most of the markdown and then gives half of it back before the close leaves a tall upper shadow and tells a weaker story, because it shows the buyers could not hold what they had taken. The hammer's specific claim is not that buyers appeared. It is that buyers appeared, took the whole excursion back, and were still holding it when the bell went. Everything else about the candle is measurement.

A hammer does not say the decline is over. It says one attempt to continue the decline was refused, and that the refusal was still standing at the close.

The measurements are not arbitrary decoration on top of the idea; each one is a way of asking whether the story above actually happened. A long lower shadow asks whether price really was pushed far below the open. A small body asks whether the reclaim was nearly total. A missing upper shadow asks whether the close held at the top rather than fading into it. Fail any of those and you are looking at a different session with a different meaning.

The vocabulary is Japanese, but the English convention has one identifiable source. Candlestick charts were largely unknown outside Japan until Steve Nison published them, first in articles in the late 1980s and then in Japanese Candlestick Charting Techniques in 1991, the book that fixed the pattern names still in use. It files the hammer and the hanging man together as umbrella lines and gives three recognition criteria: the real body sits at the upper end of the trading range, and the colour of the body is not important; the lower shadow is at least twice the height of the real body, with two to three times treated as the comfortable case; and there is no upper shadow, or only a very short one. The Japanese name recorded for the pattern is takuri, translated there as something close to trying to gauge the depth of the water by feeling for its bottom, which is a better description of the mechanism than most of what has been written since.

Two things about those criteria are worth stating carefully, because they usually get lost. The first is that the ratio is a floor and not a target. The longer the lower shadow, the shorter the upper shadow and the smaller the body, the more emphatic the rejection being recorded. A candle whose shadow is barely twice its body is the weakest specimen that still qualifies, not a typical one. The second is that these are conventions, not laws. No exchange defines a hammer. No standards body publishes a threshold. Different charting packages draw the line in different places, and a scanner that returns a tidy list of hammers is returning its author's cutoff, not a fact about the market. Treating a convention as a rule is how people end up arguing about whether something qualifies instead of asking what it recorded.

Colour is the detail people fixate on and it matters least. A dark body, where the close finished below the open, still closed near the high of the period after recovering nearly the whole markdown; the rejection happened either way. A light body is marginally stronger because the session ended at its extreme. It is a tiebreaker, not a criterion. What actually varies the information content is the depth of the shadow, the tightness of the body, the location, and the volume behind it.

Push the body down to nothing and the hammer becomes a dragonfly doji, where open and close finish at essentially the same price at the top of a long lower shadow. That is the limiting case of the same idea, and it is read the same way. Our guide to the doji candlestick works through that family and the wider discipline of treating a candle as context rather than as a signal, which is the same discipline this page applies to a shape with a body.

Here is the part that separates people who use candles from people who collect them. The anatomy tells you what happened inside the period. It does not tell you whether what happened matters. For that you need to know what the market was doing before the candle printed, and the same shape in three different places carries three genuinely different amounts of information.

After a sustained decline, a failed markdown is worth reading. Sellers have been winning for weeks; they tried again; for the first time in that stretch they were comprehensively refused, and the refusal held into the close. That is new information about the balance between supply and demand, and it is exactly the kind of thing that shows up near the end of a decline. Note the honest phrasing: it is the kind of thing that shows up there. It also shows up in the middle of declines that continue for another month.

In the middle of a sideways range, the identical candle is close to meaningless. Ranges reject their lower edge continually, because that is the definition of a range; a candle whose whole content is that price fell and came back is describing normal behaviour rather than a change in it. There is no trend to exhaust and no group of participants who have been proven wrong. The shape passes every measurement test and carries almost no information, which is a useful reminder that passing a measurement test is not the point.

In the middle of an uptrend, the same shape is not a hammer at all, and that is not a naming quibble. The reason the candle leans constructive after a decline is that the people trapped by it are the sellers underneath the close. After a rally, the people trapped are the buyers above it, and the session has just demonstrated that sellers can reach deep into the range on a market where longs are crowded. Same picture, opposite implication, and the only thing that changed was what came before.

The practical consequence is that the location test comes first. Establish that a decline happened and that price arrived somewhere participants have reason to defend, then ask whether the candle qualifies. Doing it the other way around, scanning for the shape and then looking for a story that fits it, is how a chart gets talked into saying something.

The clearest proof that location rather than shape carries the meaning is that the identical candle already has two names. Anatomy alone never names it. A small body at the top of the range with a long lower shadow is a hammer after a decline and a hanging man after a rally. Its mirror image, a small body at the bottom of the range with a long upper shadow, is an inverted hammer after a decline and a shooting star after a rally. Four names, two anatomies, and the only variable that switches the reading is position in the prevailing trend.

The logic is mechanical rather than ceremonial. Ask who is now holding a losing position because of the candle. After a decline, it is the sellers who exited along the lower shadow and are watching the market close above them; their discomfort is a source of buying if price holds. After an extended rally, it is the buyers who bought into the highs and have just watched sellers push deep into the range for the first time; their discomfort is a source of selling if price fails. The candle is identical. The inventory of trapped participants is inverted, and that is what the two names are really recording.

Classical practice treats the twins asymmetrically, and the asymmetry is instructive. Nison requires confirmation for the hanging man, typically a close below its real body, while writing that the hammer need not be confirmed. Modern practice is more conservative and confirms both, for a reason the arithmetic below makes concrete: the confirming bar is what fixes your entry price, and the entry price, measured against the hammer's low, determines the entire risk profile of the trade. Waiting is not only about being more certain. It is about knowing what you are risking.

The inverted pair mirrors the internals rather than merely the picture. An inverted hammer records a rally attempt that failed within the period: buyers pushed price well above the open and could not hold it into the close. After a decline that failure still leans mildly constructive, because an attempt to rally at all is new behaviour after days of one-way selling, but the close near the low makes it materially weaker evidence than a hammer and it asks for stronger follow-through. The same candle after a rally is a shooting star, where the failed push ran into overhead supply and the trapped traders are above the market rather than below it.

A hammer is one period's information. On its own it is a hypothesis: supply was absorbed here, and this level may hold. Confirmation is what turns that hypothesis into something you can act on, and the mechanics of confirmation are more specific than most descriptions admit.

The common standard is a following candle that closes above the hammer's real body; the conservative version wants a close above the hammer's high. Either way the operative word is closes. A bar that trades above the level during the session and then settles back inside the hammer's range has not confirmed anything. It has demonstrated that buyers could reach the level and could not hold it, which is closer to the opposite of confirmation. The distinction sounds pedantic until you have watched it happen in real time, when the touch is visible for hours and the close arrives once.

This is also where the difference between a single candle and a two-candle formation becomes clear. A hammer needs an external event, the next bar, to mean anything, which is why it is always evidence before it is a setup. Some patterns build the confirmation into their own definition instead: a bullish engulfing pattern requires the second candle to close beyond the first candle's body as part of what makes it that pattern at all, so the follow-through and the recognition happen together. Neither construction is superior. It is worth knowing which one you are looking at, because it tells you how much has already been demonstrated at the moment you notice it.

Waiting has a price and it should be stated honestly. The confirming close is usually well above the hammer's low, so the trade you can take after confirmation is a worse price than the trade you could have taken on the hammer itself. Acting on the hammer alone gets a better entry and a tighter stop; it also means acting before the evidence arrived. That trade-off is a genuine decision with costs on both sides, not a rule to be memorised, and the honest way to make it is to know what you are giving up either way rather than to pretend one side is free.

Absorption is a transaction, not a mood. For price to be marked down and bought straight back, somebody has to have bought every share the markdown sold. That is a factual claim about how much changed hands, and it is the one claim in the whole pattern that the chart can check independently, because volume is recorded separately from price.

A hammer that prints on turnover well above its recent average is saying the excursion and the reclaim were paid for in size: large supply came out, large demand met it, and demand held the close. The identical candle on thin turnover is saying much less. On a quiet tape the same shape can be produced by a small number of orders arriving in an unhelpful sequence, and the shadow records the path of a nearly empty book rather than a contested one. Nothing about the outline distinguishes the two. Only the volume pane does.

The gradient runs with liquidity and with timeframe, and both matter more than traders expect. A daily candle on a large index constituent aggregates a full session of participation from many kinds of buyer and seller, so the volume figure beneath it is a meaningful population statistic. A one-minute candle on a thinly traded counter can be shaped by a single participant working a single order, and calling the resulting shadow evidence of absorbed supply is reading an accident as an argument. Our guide to volume in trading covers how to read that pane properly, including why comparing volume to its own recent average is more informative than looking at the raw number.

Volume is corroboration rather than proof, and it deserves the same honesty as the rest of the page. Heavy turnover on a hammer is consistent with genuine absorption; it is also consistent with a large seller who has not finished. What it rules out is the least interesting case, the shape that formed because almost nobody was there. That is a narrower claim than the usual telling, and it is the one the data supports.

The long lower shadow is simultaneously the evidence and the bill. It is the evidence because its depth is what makes the rejection emphatic. It is the bill because the only honest place for a stop is below the low that the buyers defended, and the deeper that low, the wider the stop, and the wider the stop, the smaller the position for the same rupee risk. Most of the trouble people get into with this pattern is a refusal to accept that trade.

Take an illustrative case. A stock has declined for weeks and is approaching an area between 480 and 490 rupees that it broke upwards from months earlier, so there is a structural reason for resting bids to sit there. Into that area it prints the candle from the first figure: open 494, high 497, low 480, close 496. The next session closes at 502, above both the hammer's body and its high, which is confirmation on either standard. The educational entry is that confirming close at 502; the stop sits one rupee below the defended low at 479, because a trade back below that low falsifies the absorption that was the entire reason for the position; the target is 548, the underside of the shelf the decline broke down from.

Now look at what the geometry did to the position. An account risking 2,000 rupees on a setup with an eight-rupee stop could carry 250 shares. This setup allows 86, because the honest stop lives 23 rupees away from the confirmed entry. Size is the variable that absorbs the width of the stop, and it is supposed to be. Tightening the stop to afford more shares deletes the logic of the trade, because a stop placed above the hammer's low can be taken out while the pattern it is supposed to protect remains perfectly intact.

The earlier entry is a real alternative rather than a mistake, and it should be evaluated as one. Buying the hammer's own close of 496 cuts the risk per share to 17 rupees, raises the size to 117 shares, and improves the reward-to-risk ratio to roughly three to one, because the target is unchanged while the risk is smaller. It buys all of that by acting on a hypothesis that has not yet been tested. Both trades are defensible. What is not defensible is taking the earlier entry and then describing it as though the evidence had already arrived.

The shadow is the argument and the cost in the same stroke. A deeper rejection is a stronger claim and a wider stop, and the position size is where those two facts are reconciled.

The recurring failures are not failures of recognition. People identify the shape correctly and then draw more from it than one period of trading can support. Four patterns account for most of the damage.

The fourth is the one worth sitting with, because it is not a charting error at all. It is a sampling error, and it is what makes this pattern feel far more reliable than any measured record supports. A page of examples chosen after the fact will always look convincing. The only cure is to write the definition down before you go looking, including what would make you reject a candidate, and then to count the ones that did not work along with the ones that did.

Tested as a standalone trigger, the record of candlestick patterns is weak, and there is no reason to be coy about it. The most cited study, Marshall, Young and Rose in the Journal of Banking and Finance in 2006, examined candlestick signals on Dow Jones Industrial Average constituents from 1992 to 2002. The design is the part that matters: rather than compare against a naive benchmark, they generated bootstrapped random series of opens, highs, lows and closes and asked whether the signals did better on real data than on manufactured data with similar statistical properties. Single lines and multi-candle patterns alike added no statistically significant value.

The obvious objection is that candlesticks are a Japanese technique and might work best in their home market. The same authors, with Cahan, ran the test there: the hundred largest stocks on the Tokyo Stock Exchange from 1975 to 2004, published in the Review of Quantitative Finance and Accounting in 2008. They found no value across the full thirty years, none in the sub-periods, and none when bull and bear regimes were separated. That is about as direct a check as the objection deserves, and it did not rescue the patterns.

The correct conclusion is narrower than "candlesticks are useless", and getting the width right is the whole point. What those studies tested is the naked shape, bought mechanically wherever it appeared, with trend, level, volume, confirmation and position sizing all stripped out, which is to say with everything this page has spent its length on removed. They are strong evidence against the shape as a system. They are not evidence about the shape as one input among several, because that is not what was measured, and claiming otherwise in either direction would be overreading them.

What survives is the part that was never really about the candle. A defined invalidation at a specific price, a position sized from that invalidation, a target that has to clear a multiple of the risk before the trade is worth taking, and a rule about what evidence must arrive before you act. Those are useful whether or not any given hammer resolves the way you hoped, and they are what a hammer is genuinely good for: it organises a decision by giving you a low that means something. Anyone quoting a success rate for this pattern is quoting their own sample, and you should ask how the sample was collected.

Read plainly, a hammer is a well-defined piece of evidence about one period of trading: sellers reached deep, buyers took everything they offered, and the close held at the top. That is worth knowing and it is nowhere near a decision. The candle poses a question, which is whether the level that was defended will keep being defended. It does not answer it. The next bar starts to.

That framing also fixes the order of operations, which is where most of the practical value sits. The trend comes first, because it decides whether a failed markdown is meaningful or routine. The level comes second, because it explains why buyers would be there at all. The anatomy comes third, and it is a measurement rather than a judgement. Volume comes fourth, because it says whether the fight was real or the room was empty. The confirming close comes last, and it is the only part that converts a rejection into a reversal hypothesis you can act on. Building that case around a candle, rather than trading the candle, is what the method we teach is built around.

The shape takes a minute to learn. Everything that makes it worth anything takes considerably longer, and none of it is about the shape.

• Steve Nison, Japanese Candlestick Charting Techniques (New York Institute of Finance, 1991; second edition 2001). The reference text that carried candlestick vocabulary to Western markets. Source of the umbrella-line classification, the three recognition criteria for the hammer and the hanging man, the takuri etymology, and the asymmetric confirmation requirement between the twins. archive.org
• Marshall, Young and Rose (2006), Journal of Banking and Finance 30(8), 2303 to 2323. The benchmark academic test of candlestick trading strategies on Dow Jones Industrial Average component stocks, 1992 to 2002, using bootstrapped random open, high, low and close series as the benchmark; found no significant value in the patterns traded in isolation. ideas.repec.org
• Marshall, Young and Cahan (2008), Review of Quantitative Finance and Accounting 31(2), 191 to 207. The same methodology applied to the hundred largest Tokyo Stock Exchange stocks, 1975 to 2004, testing candlesticks in the market where the technique originated: no value in the full period, the sub-periods, or bull and bear regimes. link.springer.com
• Exchange open, high, low and close conventions. Indian exchanges publish official open, high, low and close values per instrument per session, and the continuous equity session runs from 09:15 to 15:30. Every candle discussed on this page is constructed from those four exchange-recorded prices and nothing else.