Most candlestick patterns are not worth memorising. The handful that earn their place — engulfing, pin bar, inside bar and doji — all describe the same thing: an attempt to continue that failed. Location matters far more than shape; the same pattern at a defended level and in mid-range are different trades.
Search for candlestick patterns and you will find charts of forty or sixty named formations, each with a success rate quoted to one decimal place and no mention of what was tested, over what period, or against what benchmark. That is the part to be sceptical of. The patterns themselves are fine. The claims made about them are not.
Start with what the research found
It is worth being honest about the evidence before listing anything. The most-cited test of candlestick charting examined Dow Jones Industrial Average component stocks from 1992 to 2002, using an extension of the bootstrap method that generates random open, high, low and close prices to give the patterns a fair comparison. The conclusion was blunt: candlestick trading strategies did not create value for DJIA stocks (Marshall, Young & Rose, "Candlestick technical trading strategies: Can they create value for investors?", Journal of Banking & Finance 30(8), 2006, pp. 2303–2323).
That result is not a reason to close the tab. It is a reason to be precise about what a pattern is for. Tested as a mechanical entry signal, in isolation, on daily bars, across a broad universe of large-cap stocks, candlestick patterns did not beat chance. Nobody serious uses them that way. They are used as one input among several — confirmation that a level you already cared about is being defended, on a trade whose risk was defined before the pattern appeared.
Read that way, the finding is useful: it tells you exactly how much weight a pattern can carry. Not zero, and nowhere near enough to be the reason you took the trade.
The four that earn their place
Almost every named formation is a variant of one of these four. Learn the four and you can name the rest on sight without memorising a catalogue.
| Pattern | What it looks like | What it is telling you |
|---|---|---|
| Engulfing | A candle whose body completely covers the previous candle's body, in the opposite direction | One side gave up an entire interval's ground in a single bar |
| Pin bar / hammer | A small body with a long wick on one side, at least twice the body's length | Price reached an extreme and was rejected before the close |
| Inside bar | A candle whose entire range sits within the previous candle's range | Compression — the argument paused, and a decision is pending |
| Doji | Open and close at effectively the same price, so almost no body | Genuine balance; whatever the interval did, it resolved nothing |
The names you did not need: a hammer is a pin bar at a low, a shooting star is a pin bar at a high, a hanging man is a hammer in an uptrend. A bullish harami is an inside bar with a colour condition. A dragonfly doji is a doji that is also a pin bar. Same four shapes, different labels.
Location is the pattern
Here is the part that pattern lists leave out, and the part that does most of the work.
A bullish engulfing bar in the middle of a range is a fact about two candles. The identical bar forming at a level that has been defended twice on the daily chart, after a move down into it, is a record of buyers stepping in exactly where they stepped in before. The shapes are indistinguishable. The trades are not remotely the same.
So the sequence is always the same, and the pattern is never first:
- Mark the level before the session. On the higher time frame, as a zone. This is the support and resistance work that everything else depends on.
- Wait for price to arrive. Not near it, at it. Most sessions this never happens, which is the correct outcome.
- Then read the candle. A rejection wick or an engulfing close at the level is confirmation that the level is being defended in real time.
- Define the trade before entering. Entry, target, and the price that proves you wrong — see why every trade needs an invalidation in writing.
Run in that order, the pattern is doing the one job it is good at: telling you the level is alive right now. Run in the reverse order — scanning for patterns and then looking for a level to justify one — you will find a justification every time, because there is always a line you can draw.
The ones to stop trading
Not every named pattern is worth the screen space. These come up constantly and rarely help:
- Three white soldiers / three black crows. Three consecutive strong closes in one direction. By the time the third one closes, you are entering after a completed run — the definition of chasing.
- Morning and evening stars. Three-candle reversal formations with several conditions that must line up at once. They occur rarely, and their looser variants occur often enough to be meaningless.
- Any pattern found by a scanner alone. A scanner has no idea whether the bar formed at a level, in thin holiday conditions, or one minute before a scheduled release. It matches geometry.
- Patterns on very low time frames. On a 1-minute chart a pin bar can be a single large order clearing the book. The lower the time frame, the more of what you are reading is noise rather than intent.
- Patterns on Heikin Ashi candles. Heikin Ashi averages across bars deliberately, so a "doji" there is a statement about a smoothed series, not about real traded prices.
How to actually use one
A worked example, with the numbers left abstract because the instrument does not change the logic. Suppose a level at 100 has been defended twice on the daily chart. Price sells off into it during the session and prints a candle with a long lower wick to 99.60 and a close back at 100.30.
What that tells you: sellers reached 99.60 and could not hold it. The level is being defended. What it does not tell you: that price is going up. The trade still has to be built.
- Invalidation goes below the wick, not below the body — the wick is the extent of the failed attempt, and placing the stop inside it means the same test that just happened would remove you.
- Position size comes from that distance and your risk per trade, never from how convincing the candle looked. That is the whole point of sizing from risk rather than conviction.
- Targets come from the next levels above, marked in advance.
- If the level fails, the pattern was wrong and the trade closes at the invalidation. A confirmed pattern that loses is not a broken system; it is one observation with a known cost.
Frequently Asked Questions
Which candlestick patterns are actually worth learning?
Four cover almost everything the rest describe: the engulfing bar, the pin bar or hammer, the inside bar and the doji. Each is a plain statement about the balance between the open and the close, and each is only meaningful where it forms. Learning forty named formations adds vocabulary, not information, because most of them are variations on these four.
Do candlestick patterns actually work?
Not as standalone signals. A study of Dow Jones Industrial Average component stocks from 1992 to 2002, using a bootstrap method that generated random open, high, low and close prices for comparison, concluded that candlestick trading strategies did not create value for investors. Patterns are best used as confirmation of a trade that levels and structure already justified, not as a reason to enter.
What is the most reliable candlestick pattern?
There is no most reliable pattern, and any source naming one without stating the sample it was tested on is guessing. Reliability comes from location rather than shape: an engulfing bar at a level that has already been defended twice is a different event from the identical bar in the middle of a range, even though the pattern scanner flags both.
How many candles does a pattern need to be valid?
One or two is usually enough, and more is usually worse. The four patterns worth knowing use one or two candles. Three-candle and five-candle formations occur far less often, require more conditions to be satisfied at once, and by the time the last candle closes the move they describe has typically already happened.
Bottom line
Learn four patterns, not forty, and treat every one of them as a comment on a level rather than a signal in its own right. The strongest published test of candlestick strategies as mechanical entries found no value in them, which is exactly what you would expect from a tool being asked to do a job it was never capable of. Used properly — closed candle, pre-marked level, risk defined first — a pattern is a useful last check before committing. Build the foundation with how to read a candlestick chart and technical analysis basics, and see how the room's trade callouts work if you want to know what that looks like in practice.