Hammer, Shooting Star, Doji and Long Candles: Meaning and Measured Results

Hammer and hanging man: candles with a long lower wick

A single candlestick pattern is named from the shape of one candle. The body spans the open (the price when the candle starts) and the close (the price when it ends), the range runs from the high to the low, and the wicks are the parts that stick out above and below the body. A hammer is a candle with a small body and a long lower wick. For the measurements in this article, we picked candles whose body was at most 35% of the range, whose lower wick was at least twice the body, and whose upper wick was at most 15% of the range. The shape means price was pushed down hard during the candle, then came back up and closed in the upper part of the range. The same shape is called a hammer when it follows a decline and a hanging man when it follows a rise, and it is common to use the same name whether the candle is bullish or bearish.

Shooting star and inverted hammer: candles with a long upper wick

A shooting star is a hammer turned upside down: a candle whose body is at most 35% of its range, whose upper wick is at least twice the body, and whose lower wick is at most 15% of the range. The shape means price rose sharply during the candle, then was pushed back down and closed in the lower part of the range. After a rise it is called a shooting star, and after a decline an inverted hammer. The inverted hammer and the hanging man were not included in this course's measurements, so the numbers below apply only to the hammer, the shooting star and the doji.

Doji and long candles: the length of the body

A doji is a candle whose open and close are almost the same, so the body looks like a line. In this article, every candle with a body of at most 10% of its range was counted as a doji, regardless of the move before it. Of 29,946 daily candles across 10 coins, just over 3,300 qualified, or about 11%. There is no fixed rule for how small the body must be, and this site's Candlestick Pattern Scanner uses less than 12%. A long candle is the opposite: its body is noticeably longer than those of recent candles and its wicks are short, meaning price moved strongly in one direction during that candle. A bullish one is called a long bullish candle and a bearish one a long bearish candle. Standards for how long is long also vary, so long candles were left out of this measurement.

Position comes before shape

As the first figure showed, the same shape takes on a different name and meaning depending on the move before it. That is why many books stress that a candle with the right shape in the wrong place should not be treated as a pattern. The trouble is that there is no single agreed way to define 'the move before it' either. In this article's measurements, a pattern counted as coming after a decline if the close of the candle just before it was lower than the close five candles earlier, and as coming after a rise if it was higher. Change that window to 10 or 20 candles and both the number of patterns found on the same chart and their results can change.

The same shape carries different weight on different timeframes

A hammer on the daily chart means that 24 hours of trading ended in that shape. A hammer on the 1-minute chart covers one minute of movement, and on a thinly traded coin a handful of orders can create one. The shorter the candle, the more often these shapes appear and the less trading each one contains. This site's Candlestick Pattern Scanner detects and displays seven patterns in real time on the 1-minute candles of five coins in the KRW market of Bithumb, a Korean exchange. Its definitions also differ slightly from this article's measurements: for example, it only looks for a hammer when the candle is bullish and the one candle before it is bearish. It finds shapes but does not show or guarantee what happens afterwards, so it is best used to watch how patterns form on real charts.

Measured on daily bars of 10 coins

We searched 29,946 daily bars of BTC, ETH, BNB, XRP, ADA, DOGE, LTC, LINK, TRX and SOL on Binance spot (from each coin's listing date to September 22, 2026; BTC from August 2017) for patterns, and compared the close of the pattern candle with the close 5 and 20 bars later. The up rate is the share of cases in which the close k bars later was higher. The comparison point is not 50% but the value measured across all bars, because coins rose so much over this period that even a day picked at random had a positive average return. For all three patterns, the up rate 5 and 20 bars later was never more than 2.2 percentage points away from the baseline. The counts below are the number of samples measured 5 bars later.

• Baseline (all bars): 50.5% after 5 bars, 50.4% after 20 bars
• 3,320 doji: 50.7% after 5 bars, 51.3% after 20 bars
• 756 hammers (after a decline): 51.5% after 5 bars, 49.2% after 20 bars
• 501 shooting stars (after a rise): 48.3% after 5 bars, 50.4% after 20 bars

BTC alone, and 4-hour bars alone

Narrow the data and the numbers swing more. On 19,928 BTC 4-hour bars (August 2017 to September 2026), the up rate after 532 hammers was 53.6% 6 bars (1 day) later and 57.9% 30 bars (5 days) later, above the baselines of 52.0% and 52.8%, while after 389 shooting stars it was 46.0% and 47.7%, below the baselines. That is the direction the textbooks describe. On BTC daily bars, however, the 5-bar up rate after 47 shooting stars was 57.4%, actually higher than the 53.1% baseline, and the 20-bar up rate after 79 hammers was 51.9%, lower than the 53.7% baseline. The direction flips depending on the coin and the bar length, and when many combinations are examined at once, a few of them will show large gaps by chance alone.

Why are the differences so small?

Several explanations are possible. A single candle is only four numbers, so it cannot tell a hammer that marks a rebound after big news from one that is just noise during a quiet hour. Another point is that the times at which candles open and close are only a convention: on an exchange with a different daily close time, the same day's candle can take a different shape. There is also the idea that because everyone knows these shapes, if a rise really did follow them often, many people would act first and the difference would shrink. Whichever explanation is right, in the combined daily data of 10 coins, a single shape did little to separate the direction that followed.

What this article does not say

These measurements do not prove that candlestick patterns are useless. We did not measure cases that combine other conditions, such as a hammer near support or a long candle on very heavy volume, and changing the definitions or the rule for the preceding move changes the numbers too. Conversely, a combination that came out above the baseline is no sign that it will keep doing so. What this data can say is that across the combined daily bars of 10 coins, the direction after a single shape on its own was almost no different from the baseline, and that when the data was narrowed, the direction was mixed. A pattern's name is a starting point for examining a chart, not a conclusion.