Candlestick patterns have the best marketing in all of technical analysis: evocative names (doji, hammer, engulfing, three black crows), centuries of pedigree, and a chart that lights up green and red like it is telling you something. The question nobody selling a candlestick course wants you to ask is the only one that matters: when you turn a pattern into an exact rule and test it, does it actually predict anything?
What candlesticks claim
The premise is that the shape of one or a few bars — where price opened, closed, and ranged — reveals the psychology of buyers and sellers, and therefore what comes next. A hammer means sellers were rejected; an engulfing bar means a reversal. It is an appealing story. But a story is not an edge, and “it reveals psychology” is not a testable claim until you define exactly what the pattern is and what you would do about it.
How to actually test a pattern
The honest test is the same one I would apply to anything: write the pattern as a precise rule, apply it the same way across many years and several markets, enter and exit mechanically, and include commissions and slippage. Then — crucially — check it on data it was never tuned on. If a candlestick pattern has a real edge, it survives that. If it only “works” on the chart you are staring at, it will not.
What usually happens
Put common single- and two-bar patterns through that wringer and most of the magic evaporates. Once you account for costs and stop cherry-picking, the majority show little to no durable edge on their own; a few display a weak, short-lived tendency in specific conditions, but nothing you would build an account on. That does not mean price action is meaningless — it means a named candle, by itself, is a thin signal dressed up as a strong one.
Why they look like they work
Two forces fool us. The first is pareidolia — our brains are pattern machines, and on a long enough chart some “perfect” hammer always precedes a bounce. The second is confirmation bias: you remember the textbook example that nailed the turn and forget the dozens that did not. Put a few candlestick rules on a chart and something will always seem to confirm them, which is exactly why eyeballing feels far more convincing than it deserves to.
A fairer use
None of this makes candlesticks worthless — it makes them an input, not a system. A pattern you have actually tested, used as one condition alongside a trend filter and a real exit, can earn a small place in a rules-based approach. Treated as a standalone buy-or-sell signal because it has a memorable name, it is just a more decorative way to guess.
This is the same trap I have written about with Fibonacci levels and pattern pareidolia — and the cure is always the same: test it before you trust it.
Price has no memory (I checked)
Years ago, while studying the original Turtle rules, I tested a seductive idea: only take a signal if the previous trade was a loser — the market “owed” you one. Serial correlation, in statistics terms. The result was as flat as a roulette table’s memory. The dice don’t remember, and neither do daily bars: the next trade had no idea what the last one did.
Candlestick lore is mostly memory claims in costume. A hammer “remembers” seller exhaustion; three black crows “remember” distribution. Write those claims as exact rules, test them across decades, and the memory mostly evaporates — what’s left is noise plus whatever broad market drift sat underneath.
Why patterns feel so convincing anyway
Humans are pattern-finding machines with no off switch — the same wiring that sees faces in clouds sees reversals in wicks, which is pareidolia doing its thing. Add survivorship of anecdotes — the textbook shows the hammer that worked, never the hundred that didn’t — and a red-and-green chart that hands you a story for every outcome, and you get a belief system that survives any amount of losing. The antidote isn’t cynicism; it’s counting. Define the pattern precisely, take every instance, compare forward returns against the baseline.
What survives testing
In my testing, what carries real information isn’t the shape of one candle but context you can quantify: where price sits relative to a long-term average, how stretched it is against recent ranges, whether volatility is expanding or contracting. Those are measurable and testable — they’re the raw material real systems are built from — and none of them needs a name from 18th-century rice markets to work.
Educational only; not investment advice. RelaxedTrader is not a registered investment advisor. Past performance is not indicative of future results.
