Two- and Three-Candle Patterns: Engulfing, Morning and Evening Stars, Soldiers and Crows

Patterns read across several candles

Two- and three-candle patterns are named by how the bodies of consecutive candles relate to each other. The questions are whether a later candle's body covers the earlier one's, whether the middle candle is small, and how far the last close has recovered. Because the conditions span several candles, these patterns are sometimes presented as more reliable than single-candle ones. The definitions in this article are exactly the criteria used for the measurements below, and 'after a decline' means that the close of the candle just before the pattern's first candle was lower than the close five candles before that. Single-candle patterns were covered in the article on single candlestick patterns.

Bullish engulfing and bearish engulfing

A bullish engulfing pattern is a bearish candle after a decline, followed by a bullish candle whose body completely covers the bearish candle's body. For the measurements, we counted cases where the bullish candle opened at or below the previous close, closed at or above the previous open, and had a longer body than the previous candle. A bearish engulfing pattern is a bullish candle after a rise whose body is covered by the body of the next, bearish candle. In crypto markets the next candle almost always opens at the previous close, so in practice most of the condition comes down to whether the close moved beyond the previous candle's open. Perhaps for that reason, on the daily bars of 10 coins, bullish engulfing (1,218) and bearish engulfing (1,225) were, after the doji, the most frequent candlestick patterns counted in this course.

Harami: engulfing turned inside out

The harami is the reverse of the engulfing pattern: after a candle with a long body comes a small body that sits inside the previous body. After a decline, following a long bearish candle, it is called a bullish harami; after a rise, following a long bullish candle, a bearish harami. It means that after a big move in one direction, the movement shrank on the next candle. The right side of the figure above is a bullish harami. The harami was not included in this course's measurements, so we will not say anything about how it performed.

Morning star and evening star

A morning star is made up of three candles: after a decline, a long bearish candle whose body is at least 60% of its range, then a small candle whose body is at most 30% of the first candle's body, and finally a bullish candle that closes above the midpoint of the first candle's body. The name likens the small middle candle to a star in the dawn sky. An evening star is the reverse: after a rise, a long bullish candle, a small candle, and then a bearish candle that closes below the midpoint of the first candle's body.

Textbook gaps and a 24-hour market

In the traditional definition of the morning star, the middle candle's body must not overlap the first candle's body; it has to sit below it, leaving a space (a gap) between the two bodies. Gaps mostly appear when a market closes and then reopens at a different price, but in crypto, which trades without a break, the next candle almost always starts at the previous candle's close. So this course's measurements and this site's Candlestick Pattern Scanner look only at the shape, without the gap condition. The name is the same, but what is being counted is slightly different from the textbook pattern.

Three white soldiers and three black crows

Three white soldiers are three bullish candles in a row with long bodies. For the measurements, we counted cases where all three bodies were at least 50% of their ranges, the closes rose one after another, and the second and third candles opened inside the previous candle's body. Three black crows are three bearish candles with long bodies whose closes fall one after another. Both patterns were counted without any condition on the preceding move. The 'white' and 'black' in the names come from candle colors, but colors differ from chart to chart (on this page, rising candles are green and falling candles red), so it is simplest to remember them as three rising candles and three falling candles.

Measured on daily bars of 10 coins

Using the same data as the single candlestick patterns article (29,946 daily bars of 10 coins on Binance, from each coin's listing date to September 22, 2026), we compared the close of each pattern's last candle with the close 5 bars later. The three patterns presented as bullish signals (bullish engulfing, morning star, three white soldiers) all had up rates slightly below the baseline, and two of the three presented as bearish signals (evening star, three black crows) came in above it. That is the opposite of the textbook direction, but most of the differences are a few percentage points. The 65.8% up rate one bar after three black crows (baseline 50.6%) stands out, but it rests on only 76 samples.

• Baseline (all bars): up 50.5% after 5 bars
• 1,216 bullish engulfing: 47.9% · 1,224 bearish engulfing: 49.4%
• 135 morning stars: 46.7% · 141 evening stars: 51.8%
• 107 three white soldiers: 48.6% · 76 three black crows: 55.3%

BTC alone: the small-sample trap

On BTC daily bars alone, the numbers change sharply. Three white soldiers were followed by a higher close 5 bars later 9 times out of 12 (75.0%), while morning stars were followed by a higher close only 4 times out of 14 (28.6%), against a BTC daily baseline of 53.1%. Across all 10 coins, the same patterns came in at 48.6% and 46.7%, and on BTC 4-hour bars the 6-bar up rate was 59.5% after 74 three white soldiers and 49.3% after 73 morning stars (baseline 52.0%). With only a dozen or so samples, chance alone can produce a skew this large. The probability of getting 9 or more heads in 12 coin tosses works out to about 7%, so if you look at many patterns at once, a few are likely to look like this.

A common misreading, and how not to jump to conclusions

Summing these numbers up as 'bullish patterns are actually followed by declines' is also jumping to a conclusion. Most of the differences are a few percentage points, and changing the coin or the bar length reversed the direction. When you hear a claim about a pattern, working through the checks below in order can keep you from drawing conclusions too early. The numbers in this article are also just a measurement of one past period; what to look for when testing a strategy on historical bars continues in the article on reading backtest results.

• How large is the sample? With a few dozen or fewer, results swing widely
• Was it compared with the all-bars baseline rather than with 0 or 50%?
• Did you look at the median and the up rate before the average?
• Does the direction hold on other coins and other bar lengths?
• Was the definition fixed before the results were seen?