Candlestick Patterns: Pin Bar, Engulfing, Doji, Where to Put the Stop, and Tests

A candlestick is the way most traders draw price. Each candle tells you where price opened during that period, how high and how low it went, and where it closed. This article defines the popular patterns with numbers you can check, explains the context that gives them meaning, shows where the Stop loss goes, and looks honestly at whether these patterns give any real edge.

Anatomy of a candle

Every candle has four prices, together called OHLC: Open, High, Low and Close. From these four you get three parts.

• Body = |Close − Open|, the span between open and close
• Upper wick = High − the higher of Open and Close
• Lower wick = the lower of Open and Close − Low
• Range = High − Low

A candle that closes above its open is bullish (sea green in the figures). One that closes below its open is bearish (coral). A long wick means price reached that level but was pushed back. A long body means one side controlled price for most of the bar.

Definitions with numbers

Books don’t agree on exact definitions. The table uses common rules that can be written as numbers. The numbers are not sacred, but you need them, because without them your eye will find a pattern in any candle it wants to.

Pattern	Rule (bullish side, mirror for bearish)
Hammer / bullish pin bar	Lower wick ≥ 2 × body and upper wick ≤ body, after a decline
Shooting star	Upper wick ≥ 2 × body and lower wick ≤ body, after a rise
Doji	Body ≤ 10% of the range
Bullish engulfing	Previous candle red, this one green, Open ≤ previous Close and Close ≥ previous Open
Inside bar	High < previous High and Low > previous Low (the previous candle is the mother bar)
Morning star	Candle 1 long red · candle 2 small, body ≤ 30% of candle 1’s body and below it · candle 3 green, closing above the midpoint of candle 1’s body

The same shape changes its name with location. A hammer shape at the top of an uptrend is called a hanging man, and a shooting-star shape at the bottom of a downtrend is an inverted hammer. Price-action traders often use a stricter pin bar: a wick of at least two thirds of the range.

A doji only says that open and close were almost the same, so the two sides fought to a draw. It has no direction by itself. It matters only after a long trend that runs into an important level, and even then you need the next candle to show the direction.

The classic Japanese definition wants the middle candle of a morning star to gap away from the first. Forex trades around the clock during the week, so a new candle almost always opens right at the previous close. The gap rule is usually dropped, leaving only that the middle body sits below the first body (above it for an evening star), as in the figure.

Context matters more than shape

A candle tells you what happened during that bar. It does not tell you where the next bar goes. What gives a pattern weight is the context around it.

1. Location. A hammer at a support where price has bounced several times is very different from a hammer in the middle of a range. Decide the location before you see the candle. See how to find levels in Support, resistance and supply/demand and Pivot points.
2. Trend. A signal in the direction of the larger trend, such as a bullish engulfing during a pullback in an uptrend, is a way to rejoin the trend. A signal against the trend is a guess at a reversal, which is harder. Check the bigger picture with multi-timeframe analysis.
3. Timeframe. One H4 candle holds four hours of trading. A long wick on it reflects real pressure far more than an M1 wick that may come from a few seconds of thin liquidity, and on small charts spread eats a larger share of the move (spread on small timeframes).
4. Size versus volatility. A pin bar much smaller than the candles around it is just noise. Compare the range with ATR. If the range is less than half the ATR, give it little weight (half the ATR is our own starting number, not a standard).

Waiting for confirmation

There are two main ways to enter after a pattern completes.

• Enter at the close. The earliest price and the best RR, but you take every false signal.
• Wait for the break. On the bullish side, place a Buy stop above the pattern’s high. If price doesn’t follow through, the order never fills, so some signals drop out. But the entry is worse, the Stop loss distance is wider and the room to the next resistance is smaller. If the break doesn’t come within 1–3 candles, cancel the order (1–3 is a starting rule, pick one and keep it fixed). See order types explained.

Confirmation does not create an edge by itself. It trades fewer trades and lower RR for signals that look cleaner, and you have to measure whether that is worth it. Our test below is an example where stricter confirmation made results worse.

Where to put the Stop loss

Place it beyond the point where the pattern is proven wrong. If price gets back there, your reason for the trade is gone. Then add a small buffer to clear the spread and normal noise.

Pattern	Stop loss, bullish side
Hammer / pin bar	Below the candle’s low (the wick tip)
Bullish engulfing	Below the lower low of the two candles
Inside bar	Below the mother bar’s low (wider, safer) or below the inside bar’s low (tighter, hit more often)
Morning star	Below the lowest low of the three candles, usually the middle one
Doji	No direction. Use the doji’s low once the next candle confirms upward

The Stop loss distance is set by the pattern’s size, not by how much you would like to lose, so adjust the lot to keep the loss per trade constant. See position size from Stop loss and the 1% risk rule.

Example: bullish engulfing after a pullback

An illustrative EURUSD H1 chart. Price rises from a 1.0795 low to a 1.0902 high, then pulls back into the 1.0822–1.0832 support zone (an area price passed through on the way up). Candle 8 is red, followed by candle 9, which is green.

OHLC of the last two candles:

Candle	Open	High	Low	Close
8 (red)	1.0845	1.0849	1.0826	1.0832
9 (green)	1.0831	1.0866	1.0824	1.0861

1. Check the engulfing rule

• Candle 8 closes below its open (1.0832 < 1.0845), so it is red ✓
• Candle 9 Open 1.0831 ≤ previous Close 1.0832 ✓
• Candle 9 Close 1.0861 ≥ previous Open 1.0845 ✓
• Candle 9’s body = 1.0861 − 1.0831 = 30 pips, 2.3 times candle 8’s body (13 pips)

2. Entry and Stop loss. Enter at the close, 1.0861. Stop loss 3 pips under the lower low of the two candles (1.0824) = 1.0821.

• Risk distance = 1.0861 − 1.0821 = 40 pips

3. Lot size. A 2,000-dollar account risking 1% = 20 dollars. One lot of EURUSD is worth about 10 dollars per pip.

• Lot = 20 ÷ (40 × 10) = 0.05 lot

4. Room to resistance. The prior high at 1.0902 is 1.0902 − 1.0861 = 41 pips away, about 1R (1R = the 40-pip risk). A 2R target at 1.0861 + 0.0080 = 1.0941 needs price to break the prior high first.

5. If you wait for the break. A Buy stop at candle 9’s high, 1.0866, makes the risk 45 pips and leaves 36 pips to the prior high, which is 0.8R. Confirmation made the RR to the first resistance worse.

Before every entry, measure the distance to the next resistance against the Stop loss distance. If it is less than 1R, even a perfect-looking pattern has little room to run.

This example says nothing about whether price goes up next. It only shows the steps. Try your own numbers in the calculators at the end of the article.

The evidence: do candlesticks give an edge?

The honest answer: as standalone signals, most of the evidence says they are close to random, especially after costs.

The research we collected shows roughly this:

• Tests of standalone candlestick patterns on daily stock data, in the US and in Thailand’s SET50, found no reliable edge. Some patterns the books call “bearish” were followed by positive average returns.
• A study on EURUSD found a statistical signal on 30- to 240-minute charts but none on daily charts, and after a cost of 1 pip per trade the results turned negative on every timeframe.
• A few patterns passed rigorous tests in specific markets, such as large Taiwanese stocks on daily charts (bullish engulfing, buy side only), but a result in one market does not carry over to another on its own.

In our backtest, we used a system that enters when price pulls back to the 61.8% Fibonacci level within a trend, on gold (XAUUSD) M15, in the MT5 Strategy Tester. Its original confirmation candle is very simple: a candle that touches the level, then closes back above it as a green candle. We swapped that for three stricter confirmations and kept everything else the same.

Confirmation	Real tick data, Jan–Jul 2026	Versus original
Original (closes back above level, green)	+8,710 dollars	100%
Pin bar (lower wick ≥ 60% of range)	−8 dollars	about 0%
Bullish engulfing	+2,529 dollars	29%
Wait for a retest of the level	+4,219 dollars	48%

For 2022–2025, tested with ticks simulated from 1-minute data (less reliable), the direction was the same: original +4,713 dollars, pin bar +592, engulfing +136 and retest +2,271. All three stricter versions did worse than the original in both periods.

What we saw is that the stricter the rule, the more trades it cuts, and this system’s profit sat in “ugly” trades that the simple confirmation caught and the textbook patterns let slip. This result covers one system, one market and one timeframe. It doesn’t mean patterns are worthless everywhere, and backtest results do not predict future results.

Candlestick patterns are useful as a timing tool and as a place to put the Stop loss at a level you already chose, not as signals that predict price by themselves.

To find out whether a pattern helps your own trading, record every signal by its numeric definition, including the ones you didn’t take, in a trading journal, and compare it with entering without the pattern. Try it on a demo account before real money.

Common mistakes

• No numeric definition. “The wick is long-ish” lets you count the patterns that won and forget the ones that lost, which is hindsight bias on charts.
• Reading the shape and ignoring the location. A hammer in the middle of a range, or an engulfing smaller than the candles around it, means almost nothing.
• Judging before the candle closes. A long wick mid-bar can be gone at the close. Always wait for the close before you count it.
• Placing the Stop loss by the money you want to risk, not by the pattern. A stop closer than the wick tip sits where the pattern is not yet wrong. Change the lot instead.
• Trading patterns on very small charts. On M1 the move per trade is short and spread takes most of it, while research suggests little edge survives costs even on larger charts.
• Adding confirmations until the trades disappear. Each extra condition has to be shown to cut more losing trades than winning ones, not added because it feels safer.