Doji Star Candlestick Pattern: A Complete Trading Guide

A doji star is a two-bar reversal signal: a tall bar in the direction of the trend, then a doji that opens with a gap away from it. After a decline (tall falling bar, gap down) it is bullish, a buy setup; after a rally (tall rising bar, gap up) it is bearish, a sell setup. It can be the start of a morning or evening star.

This guide explains how to spot and trade both versions, and what our test bench measured on 6,448 bullish and 11,023 bearish daily occurrences. On daily data, the buy trade did slightly better than the same trade on random dates in average result per trade, and the sell trade slightly worse (+0.18 R against +0.16 R for the buy trade, −0.20 R against −0.17 R for the sell trade). Neither gap is statistically significant. Our verdict for the buy trade is No clear edge; only its 10-day excess return over the market reads promising, not proven. The sell trade is rated No clear edge: it lost a little more than the same trade on random dates, and its expectancy stayed negative.

What is a doji candlestick pattern?

A doji is a bar whose opening and closing prices are nearly the same, so neither side won the session. Its variants, such as the dragonfly, gravestone or long-legged versions, differ by their shadows, and any of them can be the second bar of this setup.

Understanding the doji star pattern

In Japanese candlestick terms, a "star" is a tiny bar set apart by a gap from the tall one before it. Steve Nison, in Japanese Candlestick Charting Techniques (1991), treats this setup as a warning that the move may be ending, to be confirmed by the next session.

Bullish doji star: a potential bottom

After a downtrend: a tall black bar, then a doji whose body sits entirely under the black body. If a third, rising bar then ends well inside the first one, the three bars form a morning star.

Bearish doji star: a potential top

After an uptrend: a tall white bar, then the same kind of cross, its body entirely above the white body. A third, falling bar ending deep inside the first one completes an evening star.

The diagrams above show the median shape of the setups our bench found, with the trade we test: entry beyond the far end of the first bar, stop at the other extreme of the two bars, target at twice the risk.

How to identify a doji star on your charts

1. Prior move: the close just before the first bar is at least one ATR below the close 10 bars earlier and in the lower half of the last 40 closes (buy setup), or the mirror (sell setup).
2. First bar: tall, in the direction of that move, with a body more than 1.3 times the average of the 10 previous bodies (1.6 times on a textbook pattern).
3. Second bar: open and close almost the same.
4. Gap: the body of the second bar sits entirely beyond the body of the first one. Shadows may overlap.

This is the rule our bench applies, following the TA-Lib library (CDLDOJISTAR) with a stricter tall bar and prior trend. The gap explains a striking fact: the setup almost never shows up on currency pairs. They trade around the clock, each bar starts where the previous one ended, and a gap is rare. Crypto also trades around the clock, yet our bench counted 2,351 hourly buy setups there: our rule accepts a gap of a single tick, and on crypto that gap is often tiny. On hourly and 4-hour forex charts the setup forms almost only at the weekly open, across the weekend gap, so we do not publish those rows. On daily forex charts we found 9 buy setups since 2013, and only 4 closed trades: too few to read either the 10-bar move or the 2:1 result.

The psychology behind the setup

The first bar shows one side in control, and the gap shows that control carried over. Then the second bar shows the move stalling: after starting beyond the first bar, the session ends where it began. Nison reads it as momentum fading, and the next session decides whether the stall becomes a reversal.

How to trade the doji star pattern effectively

The trade we test waits for real confirmation. For the buy setup: a buy stop above the top of the first bar, a stop under the bottom of the setup, a target at twice the risk. The sell setup is the mirror. The move must retake the whole first bar before the entry triggers, so most setups never trade: on daily data, the buy entry triggered in 28.6% of cases, and in 70.8% the stop level broke first and the setup was canceled.

For the buy setup on daily data, winning trades took a median of 17 days to reach the target and losing ones 7 days to hit the stop (12 and 6 days for the sell setup). The two bars span a median 1.82 times the average true range (ATR) for the buy setup and 1.76 for the sell setup, so the stop is wide: size the trade on the distance from entry to stop.

Combining with support and resistance levels

A buy setup at a prior swing bottom, or a sell setup under a prior top, has a reason for the reversal beyond the bars. We have not tested this filter.

Confirming signals with technical indicators

We have not tested RSI or divergence filters. We did test the 200-day average: on daily stock data, buy setups under it reached the 2:1 target in 40.7% of cases (+0.26 R), against 33.2% (+0.05 R) above it. This breakdown is descriptive, not corrected for the number of tests.

Reliability: what our bench measured

Read the table in pairs of columns. "Moved as expected" means up 10 bars later for the buy setup, down for the sell setup. On daily data, over 1,839 trades, the buy trade reached its target 37.7% of the time against 36.7% for the same entry, stop and target on random dates: +0.18 R per trade before costs, against +0.16 R. That gap is rated No clear edge: once the clustering of setups on the same dates is taken into account, it is not statistically significant. The direction after 10 bars (56.6% against 53.9%) gives an average excess return of +0.31% over the market, rated promising: it passes a simple test but not our correction for the thousands of tests we run.

The bearish version is weaker. Its 2:1 trade is rated No clear edge: its average result per trade stayed negative (−0.20 R against −0.17 R on random dates), because the same short placed on random dates lost money too. On its own, it is not a reason to sell.

Critical differences between doji, hammer, and spinning top

• A doji has almost no real body: it shows balance.
• A hammer has a modest real body at the top and a long lower shadow, after a decline: it shows rejection of the decline.
• A spinning top has a narrow but visible open-to-close spread with shadows on both sides: it shows hesitation.

The setup needs the tall first bar and the gap. On its own, the cross-shaped bar has no proven edge in our tests: on 1,194,453 daily trades, its 2:1 breakout trade reached the target in 33.7% of cases, against 33.5% on random dates. The doji page rates that trade No clear edge.

Best timeframes for this setup

No timeframe gives the buy trade a proven result. Its 10-bar excess return reads promising on daily data; on 4-hour and hourly crypto it reads no clear edge, and the trade itself does not stand out there either. On 4-hour crypto charts, the buy trade was rated No clear edge (+0.10 R against −0.08 R per trade), a gap that is not significant. On hourly crypto, the buy trade lost slightly less than the same trade on random dates (−0.02 R against −0.09 R per trade, 30.2% of targets hit against 29.2%), and that gap cannot be told from chance. For the buy setup, weekly results look good at first sight (45.3% of 2:1 targets hit against 43.0% on random weeks, or +0.48 R against +0.44 R per trade), but that gap in average result is not statistically significant, and the row reads No clear edge; the weekly sell setup lost money (−0.38 R against −0.41 R per trade, No clear edge). The setup is rare: about once every 1,315 daily bars for the buy setup, once every 769 for the sell setup.

Real examples from the bench

• Applied Materials (AMAT), January 14, 2015, bullish: the buy stop at 21.10 triggered on January 21, with a stop at 19.85; price fell back through the stop on January 30: a 1 R loss, although AMAT was up 2.6% 10 days after the setup.
• Euro Stoxx 50, August 5, 2011, bullish: the next session the index fell to 2,285.06, through the stop level of 2,346.29, before reaching the entry at 2,538.07. The setup was canceled without a trade, and the index was down 9.1% 10 days later.
• Nasdaq Composite, January 3, 1990, bearish: the sell order at 452.7 triggered on January 10, with a stop at 461.6; the index then slid and opened at 432.9 on January 16, below the 2:1 target at 434.9, so the target order filled at that open: a gain of 2.2 R.

These three cases are daily setups from our bench, on a US stock (prices adjusted for dividends) and two stock indices. The gallery of this section, Real examples from the bench, shows more, with their outcome. Trade results are in R, the initial risk of the trade.

Common mistakes to avoid

• Trading before confirmation: the trade we test waits for the entry beyond the first bar, and most setups never get there.
• Ignoring the gap: without it, the setup is a harami cross or a plain doji.
• Expecting the bearish version to pay like the bullish one. Its expectancy stayed negative.
• Looking for it on currency pairs, where the gap almost never happens.

Risk management rules

Frequently asked questions