Hammer candlestick pattern: meaning, rules and reliability

The hammer is the best-known single-candle reversal pattern. It forms when a market that has been falling opens, sells off hard during the session, and then recovers to close near where it opened. The result is a candle with a small body at the top and a long lower shadow below, like a hammer with a short head and a long handle.

The pattern comes from the Japanese rice-trading tradition and reached Western traders mainly through Steve Nison's 1991 book Japanese Candlestick Charting Techniques. The Japanese name, takuri, is often translated as "trying to gauge the depth of the water". Price dips down to test how far buyers will let it fall, then comes back up.

This guide covers the anatomy and the precise rules, the trading logic behind it, why confirmation matters, how to tell it apart from look-alike candles, and a test of what hammers on large US stocks actually did next.

Anatomy of a hammer

Every candle has four prices: the open, high, low and close. The real body is the box between the open and the close. It is green (or hollow) when the close is above the open and red (or filled) when it is below. The thin lines above and below the body are the shadows (or wicks). They mark the session's high and low.

A hammer is defined by the proportions of those parts:

1. Small real body. The open and close are close together. The body's color matters less than its size. A green body (close above open) is considered slightly stronger, because buyers not only recovered the losses but pushed past the open.
2. Long lower shadow. The classic rule is at least twice the height of the body. It shows the session traded far below where it ended up.
3. Little or no upper shadow. The session closed at or near its high.
4. A prior decline. The candle must appear after prices have been falling. Without a downtrend there is nothing to reverse.

The exact rules Finabase uses

Finabase's chart markers and scanner use the CDLHAMMER function from TA-Lib, the open-source library many trading platforms are built on. TA-Lib judges "small" and "long" against the stock's own recent candles instead of fixed percentages, using its default candle settings:

Two differences from the textbook definition are worth knowing.

The shadow test is looser. TA-Lib requires the lower shadow to exceed the body, not to be twice its size. So some candles the scanner labels as hammers will look less dramatic than the ideal.

There is no trend test. TA-Lib checks that the body sits near the previous candle's low, so the hammer must be at or near the recent lows, but it does not measure a prior decline. The pattern detector always returns a bullish reading. Deciding whether a real downtrend came first is up to you, and our test below shows why that judgment matters.

The psychology behind the shape

Candlestick patterns are worth studying because each one records a small story about who controlled the session.

A hammer's story goes like this. The market has been falling and sellers are confident. The session opens and they push price sharply lower, often to a new low for the move. Then the selling dries up. Buyers step in at the lower prices, perhaps value buyers, short sellers taking profits, or traders who were waiting for a flush. By the close price is back near the open, and sometimes above it.

The long lower shadow is the visible record of that rejection. Sellers had control and lost it within a single session. That doesn't prove the trend has changed. It proves that, at least once, lower prices attracted enough demand to absorb all the supply. The next sessions show whether that demand persists.

Confirmation: the step most people skip

Because a hammer shows one rejected sell-off, most practitioners, Nison included, treat it as a potential reversal that needs confirmation. Common confirmations:

• The next session closes above the hammer's high. This is the most common rule and the one used in our test.
• The next session gaps up and holds the gap.
• Volume on the hammer or the confirming session is well above average, which suggests real participation instead of a quiet drift.
• Location: the hammer forms at a level that already mattered, such as prior support, a long-term moving average, or a round number.
• Momentum agrees: an oscillator like the Relative Strength Index (RSI): how to read and use it shows a bullish divergence or is turning up from oversold.

Waiting for confirmation costs something: you enter higher than the hammer's close. In exchange you skip the many hammers that are followed by another leg down.

Two real hammers: one that worked, one that didn't

NVIDIA, March 7, 2025. NVDA had fallen about 16% in ten sessions. The March 7 session traded sharply lower intraday and then recovered to close near its open, leaving a lower shadow roughly two and a half times the body. Over the following ten sessions the stock gained about 4%. This is the textbook case: a steep, clear decline, a hammer at the low, and buyers following through.

UnitedHealth, May 7, 2025. UNH also printed a hammer after a decline, about 8.6% over the prior ten sessions. Within ten sessions it was down another 22%, after the company suspended its full-year financial outlook and announced a change of CEO. No candle pattern can price in news that hasn't happened yet. A trader who waited for a close above the hammer's high would never have entered. A trader who bought the hammer with a stop below its low would have been taken out with a small, defined loss instead of a large one.

What hammers did next: a test on 18 large US stocks

How much does a hammer actually tell you? We ran TA-Lib's hammer detector over two years of daily bars (October 30, 2024 to October 29, 2026) for 18 widely held US stocks and ETFs: AAPL, AMZN, BAC, CVX, GOOGL, HD, IWM, JNJ, JPM, KO, MSFT, NVDA, PG, QQQ, SPY, UNH, WMT and XOM. For each hammer we checked whether the price was higher five trading sessions later, and compared that with every day in the sample.

Three points stand out.

A hammer alone was close to a coin flip, no better than a random day. In a period when these large stocks mostly rose, any day had a 55% chance of being higher a week later. Hammers did about the same.

"After a decline" did not help by itself. Requiring a ten-session decline before the hammer, the textbook condition, slightly lowered the hit rate. Falling stocks often keep falling, and a single candle does not change that.

Confirmation helped a little, honestly measured. If you measure from the hammer's close, confirmed hammers look excellent. But the confirming day's own gain is then counted as a result, which is circular. Measured from the confirmation close, the point where you could actually have bought, confirmed hammers beat the baseline by only a small margin.

Hammer versus its look-alikes

Several candles share the hammer's outline. Context and small details tell them apart.

The hammer and the hanging man are the same candle. Only the trend before them differs. That is the clearest demonstration that a candle pattern is never a signal without its context.

Trading a hammer: a framework

This is an educational outline of how practitioners approach the pattern. It isn't a recommendation to trade.

1. Establish the context. Is there a clear decline? Is the hammer at a level that matters, such as support, a moving average, or the low of a range?
2. Wait for confirmation. A close above the hammer's high, ideally on rising volume.
3. Define the risk first. The hammer's low is the natural invalidation point. If price closes below it, the buyers who defended that level have failed. A stop placed just below the low defines your loss before you enter.
4. Size the position from the risk. If the distance from entry to stop is $3 and you are willing to lose $300 on the idea, the position is 100 shares. Choosing the size first and the stop afterwards is backwards.
5. Plan the exit. Prior resistance, a measured multiple of the risk (for example twice the distance to the stop), or a trailing stop.

How Finabase shows hammers

• On charts. Add the Hammer pattern to any symbol's chart and Finabase marks each candle TA-Lib identifies, so you can see how the pattern behaved on that stock.
• In the Scanner. The Scanner lists the stocks where a candlestick pattern fired on the latest session. Use it to find candidates, then check the trend and location on the chart.
• In the technical gauge. Candlestick patterns feed the symbol page's technical summary. A bullish pattern within the last eight bars counts toward Buy and a bearish one toward Sell. Like every gauge input, it describes the chart. It does not forecast it.

Frequently asked questions

Does the color of a hammer's body matter?

A little. A green (bullish) body means the close finished above the open, which shows slightly more buying pressure, and many traders rank it higher. The shadow and the location matter much more than the color.

How long should the lower shadow be?

The classic guideline is at least twice the real body. Longer is generally better. TA-Lib's default only requires the shadow to exceed the body, so Finabase's markers include some borderline cases that you may choose to ignore.

Is a hammer more reliable on weekly charts?

Patterns on higher timeframes summarize more trading and are generally more meaningful than those on a 5-minute chart, but they are also rarer and the stop below the low is farther away. The same rules apply on every timeframe.

What is the difference between a hammer and a hanging man?

Nothing in the candle itself. A hammer appears after a decline and is a potential bullish reversal. A hanging man appears after an advance and warns of a potential top.

Summary

The hammer captures a meaningful moment: sellers pushed hard and buyers absorbed it all in one session. But a single candle is a small piece of evidence. Our data showed that hammers on large US stocks were close to a coin flip unless they were confirmed, and even then the edge was small. Use the hammer as a prompt to look closer. Check the trend, the level, the volume and the momentum, wait for confirmation, and always know where you would be wrong.