Belt Hold Candlestick Pattern: Bullish and Bearish Rules, Tested on Real Data

The belt hold candlestick pattern is a single long candle that opens at one extreme of its range and runs the other way, printed against the prior trend. A bullish belt hold opens at its low after a decline and closes far higher; a bearish belt hold opens at its high after a rally and closes far lower. It is read as a potential trend reversal.

This guide covers the definition, how to identify each version, how to trade it, and what actually happened next. Our test bench applied the same rule to full price history, which gave 92,237 bullish and 127,909 bearish ones on daily data alone across stocks, crypto, forex, ETFs, indices and commodities, and compared the outcome with random bars on the same assets.

What is the belt hold candlestick pattern?

The belt hold pattern is a one-candle reversal pattern in technical analysis. It has two versions, the bullish belt hold and the bearish belt hold, and both need a trend to reverse.

Definition and origin of the name

Steve Nison, who introduced Japanese candlestick charting to Western traders in Japanese Candlestick Charting Techniques (1991), describes the belt-hold line: a long candlestick that opens at its high or its low, with no shadow, or almost none, at the opening end. Its Japanese name, yorikiri, is a sumo term for pushing the opponent out of the ring while holding his belt. The image fits: from the first trade, one side grabs the market and pushes it the whole session.

In shape, the belt hold is an opening marubozu: no wick at the open, a long real body, and a wick at the close that is usually small. What turns that candlestick into a belt hold pattern is its location, after a downtrend for the bullish version, after an uptrend for the bearish one.

How our bench defines it

Our test bench uses a rule a computer can apply without judgment, taken from the open-source TA-Lib library (function CDLBELTHOLD), plus a prior trend: a body that is long compared with the prior candles, and a very short shadow at the opening end. The bullish belt hold must come after a downtrend (the close before the candle is at least one average true range below the close 10 bars earlier, and in the lower half of the last 40 closes), the bearish one after an uptrend.

How to identify the belt hold pattern

The belt hold pattern is easy to identify on a chart once you know what to check. The candle alone is not enough: the trend before it decides the name.

Characteristics of a bullish belt hold

• A downtrend before the candle.
• A long white (bullish) body, clearly larger than the recent candles.
• The candle opens at or very near its low: little or no lower shadow.
• The close is well above the open. Classic descriptions want it near the high; our rule requires only a long white body, so an upper shadow is allowed.
• The opening price can be a new low of the move, sometimes after a gap down.

Characteristics of a bearish belt hold

• An uptrend before the candle.
• A long black (bearish) body.
• The candle opens at or very near its high: little or no upper shadow.
• The close is well below the open. Classic descriptions want it near the low; our rule requires only a long black body, so a lower shadow is allowed.

A quick checklist

Check the trend first, then the opening end (no wick), then the body length against the last 10 candles, and wait for the close. A long candle without a wick at the open, in the middle of a range, is an opening marubozu, not a belt hold pattern. On the median daily occurrence our bench found, a bullish belt hold pattern spans 3.60% of price, or 1.18 times the average true range (ATR) of the asset: a normal-size session, not a giant candle. What makes it stand out is that most of the range is body.

The psychology behind the belt hold formation

A bullish belt hold candle tells the story of a session that starts at the low of a decline. Sellers have been in control, price sometimes gaps down at the open, and then buyers take over from the first trade. Price never trades below the opening level and climbs the whole session. The classic reading is a change in control: buyers who were absent during the decline are suddenly in charge.

The bearish belt hold is the mirror: the session opens at a new high of the rally, sometimes on a gap up, and sellers push price down from the first trade to the close. The top of the candle becomes a level to watch as resistance.

That reading is plausible, but it describes one session. Whether one session of control after a trend predicts the next bars is a question for data. We answer it in the reliability section below.

How to trade the bullish belt hold pattern

The classic bullish belt hold trading setup buys a break above the high of the candle, with a stop loss at its low and a target set as a multiple of the risk. This is the trading rule our bench simulates on every occurrence.

Identifying the entry point

Setting stop loss and profit targets

The natural stop loss is the low of the candle, which is also its opening price. If price trades below it, the session of buying has been erased. The stop covers the whole candle, so the risk is wide on a long one: size the position on the distance from entry to stop. Our bench sets the target at twice the risk (a 2:1 reward-to-risk ratio); a target at the next resistance level is an alternative we did not test. On daily data, trades that reached the 2:1 target took a median of 8 days; trades that were stopped took a median of 4 days. A trap worth knowing: of the daily trades that hit their stop, 24.3% had first been up 1 R.

How to trade the bearish belt hold pattern

Trading the bearish belt hold pattern uses the same setup upside down: sell a break below the low of the candle, stop loss at its high, target at twice the risk.

Spotting the reversal signal

The signal is complete at the close of the candle. Look for a long black body that opens at a new high of the rally, ideally at a known resistance level (a filter we have not tested). A sell stop just below the low of the candle confirms that sellers are still in control on the following sessions. On daily data, that entry was triggered in 78.9% of setups; 21.0% were canceled because price broke the high first.

Risk management strategies

The stop goes above the high of the candle, which is its opening price. Short trading on the stock market faces a headwind: the average stock rises over time, so a short taken on random dates loses money too. On daily data, trading the bearish belt hold pattern had an expectancy of −0.13 R per trade before costs, against −0.14 R for the same short on random dates. Keep the risk per trade small and fixed, and remember that a short can be stopped by a single gap.

Key differences between belt hold and marubozu

The two are easy to confuse, and the difference is simple:

• A full marubozu has no shadow at either end; a belt hold may keep a shadow at the closing end.
• A marubozu can appear anywhere; a belt hold needs a prior trend in the opposite direction.
• A closing marubozu has no shadow at the close; the belt hold has none at the open.

The results are close too. On daily data, white marubozu candles were higher 10 bars later 52.7% of the time against 54.2% for random bars; price was higher 53.4% of the time after a bullish belt hold pattern, against 53.5%. Neither beat chance on that 10-bar measure: the white marubozu reads Weak, below chance, and the bullish belt hold No clear edge.

Two-candle cousins exist. A bullish engulfing can have a belt hold as its second candle, and the kicking pattern is two opposite marubozu separated by a gap.

The importance of volume in confirmation

Volume above average on the pattern is read as a sign that the reversal had participation. We measured it. On daily data, a bullish belt hold pattern printed on volume above 1.5 times the 20-day average reached the 2:1 target in 37.8% of cases, against 37.0% on normal volume, with an expectancy of +0.16 R against +0.11 R. For the bearish version, high volume left the target rate flat (28.8% against 29.2%), with an expectancy of −0.10 R against −0.13 R. This breakdown is descriptive (not corrected for the number of tests): treat volume as a small difference, not a filter that changes the odds.

Reliability and limitations of the pattern

The belt hold candlestick pattern is not reliable on its own: on no market or timeframe we tested did its average excess return over the rest of the market pass our statistical test in the expected direction, and on hourly charts both versions moved against the expected direction relative to their market: the bullish version reads worse than chance (−0.06%), the bearish one weak, below chance (−0.04%). Here are the trading statistics from our test bench, which applies the same rule to full market history.

Bullish belt hold

Bearish belt hold

How to read these tables

Read each table in pairs of columns. On daily data, 53.4% of bullish belt hold patterns were followed by a gain 10 days later. That sounds bullish until you see that random bars on the same assets rose 53.5% of the time: the average stock or coin rises a little more often than it falls. The excess return over the rest of the market after 10 days is +0.01%, practically zero, with a 95% confidence interval from −0.06% to +0.09%. Zero is inside the interval: no clear edge.

The 2:1 trade columns tell a similar story. Trading the bullish belt hold pattern on daily data, the 2:1 setup reached its target 37.1% of the time, against 36.1% for the same entry, stop and target on random dates, for +0.12 R per trade before costs against +0.09 R. The gap does not pass a simple significance test (p < 0.05), let alone our correction for multiple testing: the row reads No clear edge, and we do not count it as an edge. For the bearish version, the gap in average result per trade (−0.13 R against −0.14 R, with 29.2% of targets hit against 28.8%) does not pass it either. Fees and slippage would take part of either difference.

One pattern of results is worth knowing: on hourly data, the next bar after a bullish belt hold rose only 45.8% of the time, against 48.0% for random bars, and the next bar after a bearish one fell 46.2% of the time, against 48.7%. These one-bar figures are descriptive: they are not part of the tables on this page and we did not test them for significance. They describe a one-bar pullback, not a tested effect.

Why context matters in technical analysis

We split each daily bullish belt hold pattern by the 200-day moving average. Above it (a pullback in a long-term uptrend), 53.9% were higher 10 days later; below it, 53.2%. The 2:1 trade gave +0.12 R above it and +0.12 R below it. For the bearish version, the trade gave −0.13 R above the average and −0.11 R below it. The larger trend barely changed the result. Over time, the bullish excess return stayed close to zero in both halves: it was +0.03% in the first half of the history and 0.00% in the second. These splits are descriptive, not in the tables and not corrected for the number of tests.

Common mistakes to avoid when trading belt holds

• Calling any long candlestick a belt hold. Without a prior trend in the opposite direction, it is just an opening marubozu.
• Expecting the candle to work as a trading system on its own. In our tests it showed no edge that survives our correction for multiple testing, on any market or timeframe.
• Entering at the close and expecting the move to extend right away. On hourly data, the next bar more often went against the candle (a descriptive figure, not tested for significance).
• Setting the stop inside the candle to make the risk look smaller, then sizing the position as if the risk were small.
• Holding for a big reversal. Of the daily bullish trades that hit their stop, 24.3% had first been up 1 R.

Indicators that pair well with the belt hold

Since the candle alone has no proven edge, some traders combine it with technical indicators that describe the trend or the momentum around it. Indicators do not change what the candlestick is; they add a second, independent reason to act, or not to act.

Using moving averages for trend confirmation

The 200-day moving average is the filter we tested, and it barely moved the result (see above). Some traders use shorter averages, such as the 20 or 50-day, to define the pullback: for them, a bullish belt hold candle that forms at a rising 50-day average in a larger uptrend is the setup to look for. We have not tested that version.

Incorporating RSI for overbought or oversold levels

One trading rule is to take a bullish belt hold pattern only when the RSI is oversold (below 30), and a bearish one only when it is overbought (above 70), on the idea that the prior trend is stretched. We have not tested an RSI filter: if you use one, test it on your own market before you rely on it.

Other indicators traders use

Other indicators that can be paired with the belt hold pattern are the MACD and the stochastic oscillator for momentum, support and resistance levels for location, and the ATR for sizing the stop. We have not tested these indicators: they are trading habits, not measured filters.

Belt hold patterns across different time frames

Belt hold patterns form on every market and timeframe, and no timeframe kept an edge over chance after our correction for multiple testing. The pattern is frequent: about one bullish belt hold every 92 daily bars and one bearish every 66, and on hourly charts about one bullish every 95 bars and one bearish every 98. On stocks, weekly data shows the best raw numbers for the bullish version (41.7% of 2:1 targets hit, +0.26 R), but random weeks on the same stocks did almost as well (40.7%, +0.24 R): weekly stock charts trend up, and any long trade benefits.

One subgroup that looks good is on indices. On weekly index charts, bullish belt hold trades reached the 2:1 target 52.8% of the time against 38.6% on random dates (+0.58 R against +0.16 R), on only 254 trades, and that gap passes a simple test but not our correction for multiple testing: the row reads Promising, not a confirmed edge. On daily index charts, both versions had an average 10-bar excess return below zero (−0.18% bullish, −0.11% bearish), a gap that passes a simple test but not our correction for multiple testing. On weekly charts, all markets together, the bullish version also lagged the rest of the market slightly (−0.55%), with the same unconfirmed status.

Real-world examples of belt hold setups

Here are real patterns, picked by our bench (best and worst results, textbook shapes, most recent, random draws), with their actual outcome.

A bullish belt hold that reached its target

On March 26, 2021, after a decline, DOT/USDT opened at 29.8059, dipped to 29.7894 and closed at 33.1782, near its high of 33.3689. The buy stop at 33.3689 was triggered the next day, with a stop at 29.7894 and a 2:1 target at 40.5279. Price pulled back to 31.1119 without breaking the low, then rallied and reached the target on April 2, 2021. DOT was 38.0% higher 10 days after the candle: here the reversal came fast.

A bearish belt hold that needed patience

On April 19, 2023, after a rally, DOT/USDT opened at 6.916, just below its high of 6.92, and closed at 6.309. The sell stop at 6.24 was triggered the next day, with a stop at 6.92 and a 2:1 target at 4.88. DOT was 4.2% lower 10 days after the candle, then drifted down for weeks without breaking the high (the highest point was 6.438) and reached the target on June 10, 2023.

Failed belt hold setups

On January 30, 2023, XLM/USDT printed a bearish belt hold that opened at 0.0945 and closed at 0.0899. The short at 0.0881 was triggered on February 1, with a stop at 0.0949. Price fell to 0.0817 on February 13, short of the 0.0745 target, then rallied and hit the stop on February 20: a full 1 R loss.

On August 8, 2018, Lumentum (LITE) printed a bearish belt hold that opened at its high of 61.00 and closed at 56.50. The share price never broke the 54.05 low, and on August 16 it traded above 61.00, so the setup was canceled without a trade. LITE kept rising and was 11.8% higher 10 sessions after the candle: here the cancellation rule kept the trader out of a losing short.

Tools for finding belt hold patterns

You can scan for this candlestick pattern with a charting tool that flags candlestick patterns, such as TradingView, or with the CDLBELTHOLD function of the open-source TA-Lib library. Check which definition your tools apply: thresholds for a "long" body and a "short" shadow differ, and some tools require no prior trend at all. Before trading the pattern with real money, mark the last 50 occurrences on your market, write down the entry, stop and target, and compare the outcome with the same trade on random days. That is the test we run at scale.

Frequently asked questions

Summary

• A belt hold is a long candle with no shadow at the open, against the prior trend: bullish after a decline, bearish after a rally.
• It is an opening marubozu placed at the end of a trend.
• On 92,237 daily bullish belt holds, 53.4% were higher 10 days later, against 53.5% for random bars on the same assets.
• The 2:1 trading setup on the daily bullish belt hold pattern averaged +0.12 R before costs, against +0.09 R for the same trade on random dates, a gap that does not pass our statistical test.
• Volume and the 200-day moving average changed the results only slightly, and on hourly data the next bar more often went against the candle (descriptive, not tested).

If you use this pattern in your trading, treat it as information about one session, not as a forecast. Define the stop at the opening end of the candle before entry, size the position on that distance, and keep the risk per trade small.

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 applies the same published rules to 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. This page is educational and is not investment advice: trading involves a risk of loss, and past results do not predict future ones.