Closing Marubozu Candlestick Pattern: Definition, Trading Rules and Tested Results

The closing marubozu candlestick pattern is a single candle with a long body and no wick at the end where it closes: a bullish one ends the trading day at its high, a bearish one ends at its low. Traders read it as a sign that buyers or sellers kept control of the market until the last trade.

This guide covers the definition, the two types of marubozu, how to identify the pattern on a chart and how traders use it in their trading. It also answers the question that matters for a trade: what happens after it? Our test bench applied the same rule to full price history, 370,811 white and 340,366 black ones on daily charts alone, across stocks, crypto, forex, ETFs, indices and commodities, and compared the outcome with random bars on the same assets.

What is the closing marubozu candlestick pattern?

Marubozu is a Japanese word translated as "bald" or "shaved head", which gives the meaning of the name: a candle with no shadow at one or both ends. The full marubozu has no wicks at all. This one is the partial version where only the end of the trading day is shaved, while the side of the open may keep a small wick. In technical analysis it belongs to the single-candle patterns, like the doji or the hammer.

Steve Nison, who introduced Japanese candlestick charting to Western traders in Japanese Candlestick Charting Techniques (1991), lists the marubozu among the basic candle lines. Later candlestick references treat the white and black versions of this variant as separate patterns. This page does not reuse any success rate from books: every number below comes from our own test bench.

Definition and meaning

Two features are required:

• a long real body, larger than the average body of the recent candlesticks;
• no shadow, or a tiny one, at the close: above the body (the upper shadow) for a white marubozu, below it for a black one.

Our bench follows the rule of the open-source TA-Lib library (function CDLCLOSINGMARUBOZU) with a stricter long body, written so a computer can apply it without judgment: the body is more than 1.3 times the average body of the previous 10 candles (1.6 times on a textbook candle), and the shadow at the close is less than one tenth of their average range. The shadow near the open is not constrained, so a full marubozu also meets the rule. No prior trend is required.

In the classical reading, the long body reflects strong, steady momentum and pressure in one direction, and the shaved end shows that this pressure did not fade before the close.

The bullish marubozu

The bullish marubozu of this type (also called white or green) opens, trades higher through the trading day and ends at or very near its high. It may show a little lower wick near the open. The chart below is the median shape of the clean white occurrences our bench detected, with the faded candles before it and the lines of the trading setup we test: entry above the top, stop loss below the bottom, target at twice the risk.

The bearish marubozu

The bearish marubozu of this type (black, or red on most charts) opens, trades lower and ends at or very near its low. It may show a small upper wick near the open. Its median shape and the sell setup we test are drawn below.

Anatomy of the marubozu candle

Each part of the marubozu candle is measured against the bars before it, so the same shape can qualify on one chart and not on another. Two shadows, also called wicks, frame it, and which of them is missing is what separates the variants.

Identifying the bullish version

Look for a green candle that rises from the open, a close at or within a hair of the high, and a lower shadow of any size, often small. On the median daily occurrence our bench found, the whole marubozu spans 2.94% of price, or 1.15 times the average true range (ATR) of the asset. In other words, the typical one is about as wide as an average trading day: what makes it stand out is the body, not the range.

Identifying the bearish version

The black version is the mirror: a black candle that falls from the open, a bottom at or very near the low, and an upper wick of any size. Its median daily range is 3.25% of price, or 1.16 times the ATR.

How it differs from a full marubozu

A full marubozu candlestick pattern has no shadows at either end: it opens at one extreme and ends at the opposite one. The variant on this page only needs the last-trade end to be clean. The opening marubozu is the reverse: no wick near the open, a small one at the other end. When an opening marubozu forms against the prior move, it is also called a belt hold.

One could argue that this variant is the stronger of the partial types, because the last price is the one the market accepted at the end of the trading day. Our data does not support that ranking. On daily charts, 52.8% of white ones were higher 10 bars later, against 54.1% for random bars on the same assets, practically the same result as the full white marubozu (52.7% against 54.2%). The 2:1 trading setup had an expectancy of +0.08 R per trade before costs (+0.07 R for the full one), about what the same trade earned on random dates.

A long line is a long body with short wicks at both ends. If a shaved top carried extra information, the marubozu should beat the long line. It does not: 53.2% of long white lines were higher 10 daily bars later, against 52.8% for this marubozu pattern.

Psychology behind this candlestick pattern

The white marubozu shows a trading day where buying pressure lasted until the bell. Sellers may have pushed price down after the open (the small lower wick), but buyers took over and nobody sold into the rally at the end: the last trade printed at the top. The black marubozu shows the mirror: selling pressure that never paused before the end of the period.

That is why technical traders describe the pattern as a momentum signal: everyone who held through the end of the session was on one side, and the next session may start with the same sentiment. The story is plausible, but it describes one period of price action. Whether a strong ending predicts the next moves is a question for data, and we come back to it in the reliability section.

How to identify a closing marubozu on your charts

A trader can check five points in order:

1. Wait for the bar to end. Intraday, a candle can sit at the top and then fall back in the last minutes.
2. Compare its length with the previous 10 bodies: it must be at least 1.3 times their average.
3. Check the last-trade end: for a white marubozu, the distance from the top of the body to the high should be tiny compared with recent ranges; for a black one, the distance from the bottom of the body to the low.
4. Ignore the opening side: a wick there is allowed.
5. Note the context: the trend, and any nearby support or resistance level.

Context: where the candlestick pattern matters most

The pattern can appear anywhere, which is part of its weakness: it is common. On our data it appears roughly once every 23 daily bars for the white type and once every 25 for the black type. On a watchlist of 20 stocks, that is about two a day, both colors together. Traders filter it by its location in the trend, the surrounding price action and the direction of the larger upward or downward move; apart from the prior trend and the 200-day average below, we have not tested these filters.

After a breakout: continuation signals

In the classical reading, a white marubozu ending at its high just above a trading range shows a breakout with conviction, and traders use it as a continuation signal in the direction of the break. We have not tested the trading range location.

At support or resistance: reversal signals

A marubozu against the prior trend, a white one starting from a support zone or a black one from a resistance zone, is read as a possible reversal. We measured the prior trend: on daily charts, white marubozu after an up move reached the 2:1 target 35.4% of the time, and after a down move 37.1%. The gap is small and descriptive.

After an extended run: exhaustion traps

When it appears after extended upward or downward moves, a very large candle can mark exhaustion: the last traders chasing the move. That is why the same shape is a continuation signal in one reading and a warning in another. A single candle cannot tell the two cases apart; the trend and the price level around it are the usual tie-breakers, which we have not tested.

Trading strategies for the closing marubozu

The classic trading approach follows the direction of the candle, with the entry beyond its extreme and the stop on the opposite side. This is the trading setup our bench simulates on every occurrence.

Entry rules and trade execution

Two entries are possible. The aggressive one buys (or sells) at the end of the pattern bar. The conservative one places a buy stop just above the top (a sell stop below the low for a black marubozu) and waits for a move through it. Some traders wait for confirmation instead: one of the next candles must end beyond the high. The buy stop is a simpler form of that confirmation. In our tests, the buy stop was triggered 88.8% of the time on daily charts; in 11.2% of cases, price broke the low first and the setup was canceled before any entry. The handful left expired: price touched neither level within the 20 bars the order stays active, or the data ended first.

Setting protective stop loss orders

Defining targets

The target we test is twice the risk (a 2:1 reward-to-risk ratio). On daily charts, white setups that reached that target took a median of 8 bars; those that were stopped took a median of 4. And of the white setups that hit their stop, 24.1% had first gone at least 1 R in their favor, a reminder that a partial exit or a trailing stop changes the result of a trading plan.

Filtering the pattern with secondary indicators

Indicators and filters can add confirmation to the candle. Here is what two of them did on our data, and one we have not tested.

Using volume to validate conviction

A volume filter requires heavy trading on the pattern bar. On daily charts, white marubozu printed on volume above 1.5 times the 20-day average reached the 2:1 target 35.9% of the time, against 36.2% on normal days. Heavy volume did not help, and the breakdown is descriptive, so it is not a filter that changes the odds.

Combining with moving averages

We split daily white marubozu by their place against the 200-day moving average. Above it, the 2:1 target was hit 35.8% of the time with an expectancy of +0.07 R; below it, 37.0% and +0.11 R. The long-term trend filter barely moved the outcome. This split is descriptive, not in the tables and not corrected for multiple testing.

Momentum oscillators

A momentum filter skips a bullish marubozu when a momentum indicator such as the relative strength index is already overbought, on the idea that the move is stretched. We have not tested that filter. If you use one to confirm a setup, test it on your own market first.

Quantitative performance: success rates and statistics

This marubozu candlestick pattern is not reliable on its own: on every market and time frame we tested, its average excess return over the rest of the market either did not pass our statistical test or was below chance. The following table gives the numbers from our bench.

How to read the table

Read it in pairs of columns. On daily charts, 52.8% of white marubozu were followed by a gain 10 bars later. That sounds bullish until you see that 54.1% of random bars on the same assets were too: the average stock rises a little more often than it falls. The excess return over the rest of the market after 10 days is −0.12%, with a 95% confidence interval from −0.16% to −0.08%: slightly worse than chance, a small gap that stays significant after our correction for the thousands of tests we run (the Benjamini-Hochberg procedure, with a false discovery rate of 5%): the row is rated worse than chance, and so is the black one. Confidence intervals come from a bootstrap that resamples blocks of 13 weeks or more, because on a crash day hundreds of assets print the same candle and should not count as independent evidence.

For a black marubozu, "moved as expected" means price was lower 10 bars later: 44.0% of the time, against 46.0% for random bars. The 2:1 sell setup averaged −0.14 R per trade before costs, against −0.12 R on random dates. The last column is the average result per trade in units of risk: +0.07 R means a gain of 7% of the amount risked, before fees and slippage.

Where the pattern did worse than chance

On hourly forex, both versions did worse than chance after our correction: the white setup averaged −0.05 R against 0.00 R for random dates, the black one −0.12 R against −0.01 R. On 4-hour forex, the black setup also did worse than chance, a gap confirmed by the same correction: −0.09 R against −0.02 R on 16,531 trades. On hourly crypto, the 10-bar excess return was also below chance for both colors (−0.08% for white ones, −0.09% for black ones), both rated worse than chance. On hourly charts in general, the following bar went against the candle more often than after random bars: a white marubozu was followed by a higher bar 42.7% of the time, against 48.0% for random bars. That one-bar figure is descriptive: it is not part of the tables on this page and we did not test it for significance. Read it as a hint that a strong ending on one hour was more often followed by a pullback than by more of the same.

Stability over time

The sign of the result did not change with the period: for daily white marubozu, the excess return was −0.22% in the first half of the history and −0.02% in the second half.

Best time frames and financial markets

No time frame stood out, for day trading or for swing trading. Weekly white marubozu had the best raw numbers across all markets (39.6% of 2:1 targets hit, +0.19 R), but random weeks on the same assets did slightly better (41.0%, +0.23 R): the weekly result comes from the long-term rise of prices, most of them stocks, not from the candle. On daily indices, the white setup hit the target 41.3% of the time against 39.5% for random dates; that gap in the average result per trade does not pass our test, and the 10-bar move there was worse than chance. On 4-hour crypto, 34.8% against 33.4% (+0.05 R against +0.01 R), also within noise. Forex hourly was rated worse than chance for both colors.

Closing marubozu vs engulfing patterns

An engulfing pattern needs two candles: the second body covers the whole body of the first, in the other color, after a trend. The marubozu candlestick pattern is a single candle with no trend requirement. Both indicate one side taking control. On daily charts, the bullish engulfing setup reached its 2:1 target 36.9% of the time against 35.7% for random dates (+0.12 R against +0.08 R), a gap rated no clear edge: it does not pass our test. The white marubozu did 36.2% against 36.5%. Neither gives a proven edge on its own.

Case studies: real examples

The galleries in this section show real occurrences, picked by our bench (best and worst results, textbook shapes, most recent, random draws), with their actual outcome. The case studies that follow come from daily charts, all on crypto except the first one, on the Canadian S&P/TSX index.

Example 1: a bearish marubozu that reached its target

On August 7, 2018, the S&P/TSX printed a black marubozu: open 16,491.6, high 16,494.1, low 16,281.0, ending at 16,286.3. The sell stop at 16,281.0 was triggered the next day, with a stop at 16,494.1 and a 2:1 target at 15,854.8. Ten sessions after the candle the index was 0.1% higher, and on August 27 it came back to 16,475.3, close to the stop. The target was only reached on October 9, 2018: the trade worked, the 10-bar direction did not.

Example 2: a stop and a canceled setup

On February 15, 2023, AAVE/USDT printed a white marubozu from 83.1 to 89.5, topping at 89.6. The entry triggered the next day. Price rose to 95.5 on February 19, about 0.8 R in gain, then fell back and hit the stop at 82.2 on February 22: a full 1 R loss, and AAVE was 10.6% lower 10 days later. On November 21, 2023, FIL/USDT printed a black one ending at 4.192 with a low at 4.15. Price stayed above the low of 4.15 for four days, then broke the high at 4.742 on November 25: the setup was canceled with no entry. FIL was 7.5% higher 10 days after the candle.

Example 3: two bearish marubozu that hit their stop

On August 8, 2021, XLM/USDT printed a black marubozu: open 0.3074, high 0.31365, low 0.28552, ending at 0.28716. The sell stop at 0.28552 triggered the next day, with a stop at 0.31365 and a target at 0.22926. Price rallied instead and hit the stop on August 10, and XLM was 18.9% higher 10 days after the candle. On August 20, 2018, ADA/USDT printed another one (0.10131 down to 0.09174). The entry at 0.09159 triggered the next day, price dipped to 0.08682 on August 22, then rallied and hit the stop at 0.10299 on August 28.

Scanning for the pattern with trading software

Technical analysis software can scan for candlestick patterns. The open-source TA-Lib library provides a CDLCLOSINGMARUBOZU function, and charting tools with a pattern screener can flag marubozu candlesticks. Check which definition your tool uses: thresholds for a "long" body and a "very short" shadow differ, and so do the results. Before trading the pattern with actual money, mark the last 50 occurrences on your market, write down the entry, stop and target you would have used, and compare the outcome with the same setup on random days.

Common mistakes to avoid when trading this candlestick pattern

• Acting before the bar is complete: a candle at the top with ten minutes left is not the pattern yet.
• Treating it as a reason to buy by itself: on its own, it showed no proven edge in our data, and its 10-bar move did slightly worse than chance on daily charts.
• Ignoring the size of the stop: a large body means a wide stop and a smaller position.
• Buying after an extended run without a plan for a pullback.
• Expecting a busy day to improve the odds: it barely changed the results.
• Remembering only the winners: charts make the good examples easy to find afterwards.

Advantages and limitations

Advantages

• Easy to spot: one candle, two conditions.
• Clear invalidation point at the opposite extreme.
• It forms on every market and time frame.

Limitations and false signals

• No market or time frame beat random bars: the few rows that pass a simple test without surviving our correction for multiple testing all lean below chance.
• On hourly charts, the next bar more often went the other way, a descriptive one-bar figure that is not in the tables and was not tested for significance.
• The stop is wide, and 11.2% of daily white setups were canceled before entry.
• It forms so frequently that, without filters, it produces many false signals.

Managing risk with the marubozu

If you trade it, treat the candle as a location for a position, not as a reason for one. Define the stop beyond the far extreme before entry, size it on that distance, and keep the risk small. The median losing setup on daily charts was stopped within 4 bars. Record every setup, winners and losers, and compare your trading results with the same rules applied on random dates: that is the only way to know whether your filters help your decisions.

Frequently asked questions

Summary of key points

• This marubozu is a long candle with no wick at the end where it closes: at the high for the white type, at the low for the black type.
• On 370,811 daily white marubozu, 52.8% were higher 10 bars later, against 54.1% for random bars on the same assets.
• The 2:1 setup on daily white marubozu had an expectancy of +0.08 R before costs, against +0.09 R on random dates; on black ones, −0.14 R against −0.12 R.
• On hourly forex, both types did worse than chance after correction, and so did the black type on 4-hour forex.
• Volume and the 200-day moving average barely changed the results.

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. 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. The full methodology explains every step.