Closing marubozu: a candle that ends at its extreme, tested

A closing marubozu is a long candle that finishes exactly where the action was: a green one closes on its high, a red one closes on its low. Unlike a full Marubozu candlestick: the shadowless candle, rules and real data, it is allowed a shadow at the other end. The session may have opened, wobbled against the trend and left a tail behind, as long as the final print lands at the extreme.

Many traders consider the close the most informative price of a session. It is the level at which participants were willing to carry a position overnight, and in US equities it is when a large share of institutional and index volume trades. A candle that closes at its high shows buyers still lifting offers in the final minutes. A candle that closes at its low shows nobody willing to step in. The closing marubozu isolates that one feature.

The trouble, which this page is candid about, is frequency. TA-Lib's test is loose enough that the pattern fired on 3,798 of 29,760 daily bars in our sample, roughly one day in eight. A signal that common is close to a description of ordinary price action. Below are the exact rules, two real cases from early 2025, and what those 3,798 candles did next.

What the candle looks like

White (green) closing marubozu. The candle has a long body, the close is at or very near the high, and there may be a lower shadow under the open. The story: sellers had a moment after the open, buyers took over, and they were still in control at the bell.

Black (red) closing marubozu. A long body with the close at or near the low, possibly with an upper shadow above the open. Buyers may have pushed early, but sellers held the close.

Nison groups this candle with the opening marubozu, its counterpart that has no shadow at the open. In TA-Lib's catalog the opening version goes by the name Belt hold candlestick: opening at the extreme, rules and data. The two are easy to mix up, so it helps to anchor the names on which end is clean: closing marubozu, clean close; belt hold, clean open.

The exact rules Finabase uses

Finabase marks this pattern with TA-Lib's CDLCLOSINGMARUBOZU on daily bars:

The output is +100 for a qualifying green candle and −100 for a red one. A candle with an equal open and close counts as green in TA-Lib, but such a candle cannot pass the long-body test, so it never matters here.

How loose is loose?

Two numbers explain the frequency:

1. "Long" means longer than average. In our sample, 41% of all daily candles had a body longer than the average of the ten before them. The body test alone passes about two days in five.
2. The closing-end shadow only has to be under a tenth of the average range. Strong closes at the high or low are a routine feature of trending days.

Combine them, and one test out of three is all that stands between an ordinary above-average day and a pattern marker. In our sample:

Why the close end matters to practitioners

The case for paying attention to closing strength is real, even if the pattern label is overused. A close at the high means the last buyers of the day paid the top price, and nobody sold into them enough to leave a shadow. If that happens after a series of down days, it can be the first sign that the balance has changed. If it happens in an uptrend, it shows that holders were comfortable taking the position home.

A close at the low is the opposite: the final trades hit every available bid. After a rally, that is the kind of close that shows profit-taking accelerating rather than fading into the bell.

The shadow at the opening end adds nuance. A white closing marubozu with a long lower shadow is a session that started badly and was reversed, a combination of a Hammer candlestick pattern: meaning, rules and reliability-like rejection and a strong close. A black one with a long upper shadow had a failed rally before the decline took over.

Confirmation

Because the candle is so common, the follow-through matters more than the marker. Typical checks:

• A next-day close beyond the extreme: above the high after a green candle, below the low after a red one. 993 of 2,049 green (48%) and 728 of 1,749 red (42%) closing marubozus in our sample did that.
• No immediate reversal of the close: if the next session opens with a gap against the candle and holds it, the strong close was not carried forward.
• Location: a strong close through a level that matters, such as the top of a range, is more interesting than one in the middle of nowhere.

Two real closing marubozus

Intel, February 21, 2025. INTC had climbed 28.3% in ten sessions. On February 21 it opened at $25.96, rallied to $26.29, and then fell all day to close at $24.87, three cents above the low. The upper shadow ($0.33) was too long for a full marubozu, but the lower shadow was well under the 15-cent limit, so TA-Lib called it a closing marubozu. The next session closed at $24.27, below the low, confirming it. Intel was down 4.6% after five sessions and 17.0% after ten.

Tesla, February 28, 2025. TSLA had dropped 17.7% in ten sessions. On February 28 it opened at $279.50, slid to $273.60, then rallied to close at $292.98, 90 cents under the high, inside TA-Lib's $1.75 tolerance. That is a white closing marubozu with a $5.90 lower shadow, the kind of reversal-looking day that tempts buyers. The next session opened higher at $300.34 but closed at $284.65, failing to confirm. The decline resumed: −10.3% after five sessions and −14.7% after ten.

What 3,798 closing marubozus did next

We ran CDLCLOSINGMARUBOZU over two years of daily bars (October 30, 2024 to October 29, 2026) for 60 large US stocks and ETFs, and checked five-session outcomes in the candle's direction.

Confirmed figures are measured from the confirmation close. Returns are raw price changes: after red candles the average was still positive.

We also split out the closing marubozus that were not full marubozus, the ones with a noticeable shadow at the open, to see whether the clean close carried information on its own. Using the same detections, 53.0% of those 1,366 green candles were higher after five sessions, and 48.0% of the 1,183 red ones were lower. Both are within about one point of a random day.

The reading is straightforward:

• Close to random on both sides. Green candles beat the baseline by 0.7 points, red ones by 1.1 points. With samples this large, the numbers are precise, and what they show is a tiny effect.
• Confirmation helped the green side a little. 57.6% of confirmed green candles were higher, about four and a half points above any day, a pattern consistent with short-term momentum. Confirmed red candles showed no edge.
• Most formed with the trend. Only 32.8% of green ones followed a 10-session decline and 36.7% of red ones a 10-session rise.

Closing marubozu and related candles

An educational outline

For learning purposes only, not a recommendation.

1. Ignore the marker on its own. Ask whether the strong close happened at a meaningful level, after a meaningful move, on meaningful volume.
2. Look at the opening shadow. A long tail at the open on a green candle means the session rejected lower prices before closing at the top, which some traders weigh more heavily.
3. Use the close as a line in the sand. If the next session closes back below a green closing marubozu's midpoint, the strong close did not carry over.
4. Illustrative numbers: green closing marubozu from $70.00 (low) to $74.00 (close and high). Entry on a next-day close of $74.50, invalidation under $70.00, risk about $4.60 per share; a $460 budget is 100 shares.

How Finabase shows closing marubozus

• Chart markers. Turn on Closing Marubozu (the Tesla link above opens it). Expect a dense scatter of markers, roughly one bar in eight, and overlap with Marubozu, Belt Hold and Long Line Candle markers.
• Scanner. Green detections go to the Bullish list and red to the Bearish list, following the sign of TA-Lib's output. On a typical day many S&P 500 names qualify.
• Signals tab. In the Candlestick gauge, the most recent closing marubozu within eight bars casts a Buy or Sell vote by color. In our sample that vote was Buy on 36.6% of days, Sell on 30.7%, and neutral only 32.7% of the time. Because it is almost always voting, it nudges the gauge toward the recent direction of strong closes rather than toward anything predictive.

Frequently asked questions

What does a closing marubozu mean?

That the session closed at its extreme, high for a green candle and low for a red one, with a body longer than recent average. It shows who controlled the close, not necessarily who will control the next session.

What is the difference between a marubozu and a closing marubozu?

A marubozu has very short shadows at both ends. A closing marubozu needs only the closing end to be clean. All marubozus are closing marubozus; about two-thirds of closing marubozus in our sample were not marubozus.

Is the closing marubozu the same as a belt hold?

No. They are opposite halves. The closing marubozu has a clean close; the belt hold has a clean open. A candle with both is a full marubozu.

Why does Finabase show so many of them?

Because TA-Lib's test only needs an above-average body and a close within a tenth of the average range of the extreme. In our sample that described 12.8% of all daily bars.

Summary

The closing marubozu captures a strong close: a long body that ends at its high or low. TA-Lib tests only body length and the closing-end shadow, which makes it frequent, about one day in eight across 60 large US stocks. Its five-session outcomes were within about a point of a random day, with a small momentum tilt for confirmed green candles. Treat it as a description of who held the close, and look to location, trend and the next session for meaning.