Two Crows Candlestick Pattern: Rules, Trading and Test Results

The two crows candlestick pattern is a three-session bearish reversal pattern that forms after a rising trend: a long white candle, a black session that opens with a gap above it, then a black candle that opens inside the first crow and closes inside the white body. The window left by the gap is filled, and the uptrend stalls.

This guide covers the definition, how traders use the signal, and what actually happened after it. Our bench applied the same rule to full market history, which gave 300 occurrences on daily charts across stocks, ETFs, indices and commodities, and compared the outcome with random bars.

What is the two crows candlestick pattern?

The two crows is one of the rarer candlestick patterns. The name comes from the two dark candles that "perch" above the first one, like crows on a branch, a bird often read as a bad omen. In technical analysis it belongs to the family of crow patterns and of Japanese candlesticks with three sessions, with the three black crows and the identical three crows.

Steve Nison, who introduced Japanese charting to Western traders in Japanese Candlestick Charting Techniques (1991), presents the crows as signs of a top and calls a gap a "window". The open-source TA-Lib library codes the formation as CDL2CROWS with the same construction.

Our test bench uses that rule, written so a computer can apply it without judgment: after a rising trend (the last close before the formation is at least one average true range (ATR) above the close 10 bars earlier, and in the upper half of the 40 previous closes), a long white session; a black body above it, with a window in between; another one that starts inside the middle one and ends inside the first. "Long" is measured against the 10 prior bars, as in TA-Lib.

Visual representation of the formation

The chart below is not drawn by hand. It is the median shape of the two crows our bench detected (the caption gives the count), with the faded bars showing the rise before it, and the lines of the position we test: a sell stop below the low of the formation, a protective stop above its top, a target at twice the risk.

Identification checklist

Check these points in order. All of them must be true.

1. A rising trend before the formation.
2. Day 1: a white candle that extends the advance.
3. Day 2: a down session that jumps above day 1 and ends below where it started, its body still above the first one.
4. Day 3: another down session that starts inside day 2 and ends inside the real body of day 1, which fills the window.
5. Wait for the end of the third session: before it, the formation does not exist.

Market psychology behind the two crows pattern

The formation tells a story of three days. The first session confirms the advance. The jump the next morning means buyers are still eager, but the dark session proves they could not hold that level: supply met the demand at the higher level. On the third session, the market opens near the prior close and sells off again, this time below the end of the first one. Buyers who paid up on the jump now hold a loss, which is why classic candlestick analysis reads it as a warning of a bearish trend reversal.

That reading is plausible, but it describes three sessions. In our statistics, the median formation covered 5.16% of price from top to bottom, or 2.01 times the average true range (ATR). Whether the selling continues is a question for data, not for the story.

How to distinguish the two crows from similar patterns

The table below compares the two crows with the rest of the crow family, the upside gap two crows, the three black crows and the identical three crows, and with the evening star, on the same bench, with the same rule for the position (sell stop below the formation, stop above it, target at twice the risk) and the same random baseline:

None of them beat random dates by a measurable margin, either in average result per trade on the 2:1 trade or in average excess return 10 bars later; the upside gap two crows has too few closed trades for its 2:1 trade to be judged at all.

Two crows vs. upside gap two crows

The first two sessions are the same. The difference is the last one. In the sister pattern, the third candle starts above the middle one, engulfs it and still ends above the close of day 1: the window stays open. Here, it ends inside the real body of day 1: the window is filled. The sister pattern is rarer, and its sell stop was filled less often (35.3% of cases against 63.0%).

Two crows vs. three black crows

The three black crows is three long black candles after a rise, each opening inside the prior body and closing lower, with short lower shadows and no window. It reflects three days of steady selling instead of a failed jump. In our bench, the identical three crows changes two of these rules: each crow must open at the prior close instead of inside the prior body, and its lower shadows are not limited (other definitions, such as TA-Lib's, also require them to be very short). In markets that trade around the clock, each candle opens almost exactly at the prior close anyway, which is why our bench finds more identical three crows than three black crows across all timeframes, though daily counts are close (420 against 493).

Evening star and dark cloud cover

The evening star also starts with a strong up session and a jump, but its middle body is short, of either color, and the last one ends at least 30% into the first. A version with a small middle body and a deep last session looks very much like it. The dark cloud cover does it in two sessions: a black candle that starts above the prior top and ends below the middle of the first.

Trading strategies using the two crows pattern

The classic strategy sells a break below the formation, places a stop above its top and aims for a target set as a multiple of the risk. This is the position our bench simulates on every occurrence.

Entry points and confirmation signals

The aggressive entry sells at the end of the third session. The conservative one waits for confirmation: a sell stop just below the low of the formation, which is the bottom of day 1 unless a later session trades lower. In our tests, that order was filled in 63.0% of cases; in the other 34.3%, price rose above the top of the formation first and the order was canceled before entry. In the few remaining cases, the order was never filled: neither level was touched within the 20 bars the order stays active, or the data ended first. Waiting cancels many setups: about a third of the occurrences never become a position, and we have not measured whether the canceled ones would have failed.

Stop loss and profit target techniques

Volume and the long-term trend

We split occurrences by activity and by the 200-day moving average. Formations with heavy volume (above 1.5 times the 20-day average) reached the 2:1 target in 34.8% of trades (expectancy +0.05 R, on only 23 trades, too few to read), against 26.1% (−0.19 R) for the others. Above the 200-day average, the target was hit in 27.3% of trades (−0.16 R); below it, 25.6% (−0.20 R). These breakdowns are descriptive, not corrected for the number of tests, and none of the groups had an expectancy clearly above zero.

How reliable are the two crows?

The two crows candlestick pattern is not reliable on its own: across every market and timeframe with enough history, its average excess return over the rest of the market did not pass our statistical test, and no gap survives our correction for multiple testing. Here are the numbers from our bench, which applies the same rule to full price history and compares every occurrence it finds with random bars of the same assets over the same period.

Read the table in pairs of columns. In the first row, 53.0% of two crows were followed by a lower close 10 days later; random days on the same assets were lower in 45.1% of cases. The excess return over the rest of the market after 10 days is +0.59%, with a 95% confidence interval from −0.12% to +1.20%. Zero is inside the interval: no clear edge. The excess return was +0.51% in the first half of the history and +0.67% in the second, positive in both halves, but with an interval too wide to tell it from zero.

The last three columns tell the same story. The sell position reached its 2:1 target in 27.1% of the 188 trades, against 23.3% for the same entry, stop and target on random dates, and both lose money before costs (−0.16 R and −0.30 R). Most of the loss comes from selling assets that tend to rise, which is why even random sell positions lose money. On ETFs the formation was rated Not enough data: its 14 occurrences are shown in the table but are too few to interpret.

Academic research points the same way: in "Candlestick technical trading strategies: Can they create value for investors?" (Journal of Banking & Finance, 2006), Ben Marshall, Martin Young and Lawrence Rose tested these trading strategies on large US stocks and found that they did not create value.

Why it is absent from crypto and forex

The formation needs a window between two bodies. Markets that run around the clock rarely jump: a new crypto day opens almost exactly where the prior one closed, since trading never stops. Our bench found no occurrence in all the crypto history we test, on any timeframe, and no forex occurrence at all. On stocks, where each session starts after a night of news, about once every 26,616 days.

Does the timeframe matter?

Only daily charts give enough occurrences to read. On weekly charts our bench finds 16 formations, all on US stocks, rated not enough data: their figures are not interpreted. The intraday charts we test, crypto and forex, give none.

What our test does not cover

Our numbers measure one rule and one exit, before fees, spreads and slippage, which would make every expectancy above worse. The stock sample covers the current members of the S&P 500, 400 and 600 and part of their former members, with prices adjusted for splits and dividends, and only counts occurrences with at least 5 million USD of average volume per day, so some delisted companies are included and illiquid ones are not. We did not test filters such as resistance levels, the RSI or a confirmation session, and a filter that works on your market is still possible. What the numbers do show is that the three sessions alone carry no measurable information about what comes next.

A real example of the two crows

Because the formation never printed on daily crypto charts in our data, this case comes from US stocks, on adjusted prices. On September 29, 2014, Jack in the Box (JACK) printed a white session from 53.19 to 55.00. The next session started at 55.95, above it, and ended at 55.41. On October 1, day 3 started at 55.48, inside the second body, and closed at 54.48, inside the first body. The sell stop sat at the bottom of the formation (53.07), with a protective stop at 55.96. The sell stop was never reached: on October 3 price traded up to 56.17, above the top of the formation, which canceled the order. Ten days after the signal, JACK was 2.0% lower: the decline came, but without a position. This is the 34.3% case from our statistics.

How to find the two crows on candlestick charts

Charting platforms with a candlestick scanner can flag it, and the TA-Lib function CDL2CROWS codes it. Thresholds for body size and for the window differ between tools, so the same chart can show the signal in one tool and not in another. To detect it programmatically, write the checklist as conditions on the open, high, low and closing price of three bars, with size measured against prior bars.

Best practices and risk management

• Define the protective stop before entry, and size the position on that distance.
• Keep the risk per position small: of our sell positions that hit their stop, 24.0% had first been up 1 R.
• Treat it as a location for a position, not as a reason for one: combine it with resistance, the larger trend and your own tested strategy.

Limitations of the formation

• On its own, its average excess return was not significant on any market and timeframe with enough history.
• It is rare: about once every 25,959 daily bars on a given asset, and never on crypto or forex charts in our data.
• Selling assets that rise over time has a cost: even random sell positions lost money in our tests.

Frequently asked questions

Summary of key points

• The two crows is a long white candle, a black candle opening above it, and a second crow that ends inside the first candle, after an uptrend.
• On 300 daily occurrences, 53.0% were lower 10 days later, against 45.1% for random days on the same assets.
• The 2:1 trade reached its target in 27.1% of the 188 trades, against 23.3% on random dates, with a negative expectancy for both.
• The sell stop was filled in only 63.0% of cases.
• The formation needs a window, so our bench found no occurrence on crypto or forex charts.

Past performance, including ours, does not guarantee future results, and this page is educational content, not investment advice.

About the author and the data. Written by Amaury, founder of PatternsWizard, who is passionate about finance, quantitative trading and data analysis. He built the PatternsWizard test bench, which detects each setup with fixed, published rules on full market history: crypto, forex, US stocks, ETFs, indices and commodities. The detection rules and the statistical tests are documented, and trades and win rates are compared with random bars on the same assets, excess returns with the rest of the market, with error margins and a correction for multiple testing, and the whole pipeline is checked on random-walk prices, where nothing can be predicted. Statistics on this page were generated on November 5, 2026 with version c2a1683f of the bench code. The full methodology explains every step.