Candlestick Chart Guide — Anatomy and Honest Pattern Reliability

Direct answer

A candlestick shows four prices for one interval. The body spans open to close; the wicks reach to the high and low. A body coloured one way means the close was above the open, the other way means below. Candlestick patterns are named shapes traders associate with continuation or reversal — but their predictive value is far weaker and far more context-dependent than most guides imply.

Anatomy

        │  ← upper wick (high)
       ┌┴┐
       │ │ ← body (open to close)
       └┬┘
        │  ← lower wick (low)

Body — the open-to-close range. Long body: price moved decisively between the two. Short body: it ended near where it started.
Wicks (shadows) — how far price reached beyond the body before returning.
Colour/fill — direction only. Close above open, or below.

A candle does not show the path. A long upper wick means price traded up there at some point and did not close there. It does not tell you when in the interval, or how many times.

What the shapes are said to mean

Shape	Appearance	Conventional reading
Doji	Tiny body, wicks both sides	Indecision; buyers and sellers balanced
Hammer	Small body, long lower wick	Selling pressure rejected
Shooting star	Small body, long upper wick	Buying pressure rejected
Marubozu	Long body, almost no wick	One-sided interval
Engulfing	Body fully covering the previous one	Momentum shift

These are descriptions of what happened, and they are accurate as descriptions. The dispute is entirely about what they predict.

The reliability question, honestly

This is the section most guides omit, and it is the most useful one.

What is well supported: a candle accurately summarises what traded. A long lower wick genuinely means price went there and came back — that is a fact, not an interpretation.

What is weakly supported: that any named pattern reliably predicts what happens next. Published studies of candlestick patterns report results ranging from marginal edges to none at all, with outcomes highly sensitive to market, timeframe, and the precise rules used to identify a pattern. There is no broadly reproduced finding that a given pattern works across markets and periods.

Why claimed win rates should be treated cautiously:

Pattern definitions vary. "Hammer" is not a precise specification — how long must the wick be relative to the body? Different thresholds produce different results from identical data.
Backtests are easy to overfit. Test enough variants and some will look excellent by chance.
Survivorship and selection. Examples in guides are chosen because they worked.
Timeframe changes everything. The same data resampled to a different interval produces different patterns.
Crypto is thin. On low-trade-count candles a "pattern" may be four transactions. See OHLC explained.

Anyone quoting a precise reliability figure — "the hammer is 60% accurate" — should be asked: on what market, what timeframe, what definition, over what period, and out of sample?

CoinDock does not publish pattern win rates, because we have not run that research and repeating someone else's uncited figure would be worse than saying nothing.

What candles are genuinely useful for

Dropping prediction, they remain informative:

Seeing volatility at a glance. Body and wick sizes show how much range an interval covered.
Identifying where price was rejected. Repeated long wicks at a level are a real observation about where trading stopped.
Comparing intervals. Relative body sizes show whether movement is accelerating.
Spotting data problems. A single enormous wick on an otherwise quiet chart usually means a thin book and one aggressive order — not a market event. That is diagnostic information about liquidity.

That last use is underrated and it is the one CoinDock's data supports directly: check trades_count on the anomalous candle.

Timeframe changes the picture

The same market resampled to a different interval produces different candles and therefore different patterns.

An hourly hammer may be invisible on a daily chart, and a daily doji may be a wide, noisy hour on the hourly. Neither is more real. They are different summaries of identical trades.

This is why "the pattern appeared" is an incomplete statement without naming the timeframe — and why searching enough timeframes will always find one.

Common mistakes

Treating a pattern as a prediction rather than a description.
Trusting patterns on thin markets where a candle is a handful of trades.
Reading unclosed candles. The shape can change completely before the interval ends.
Switching timeframes until a pattern appears.
Accepting a win-rate figure without a definition, market, period and out-of-sample test.

Educational content. Not trading advice, and not a signal service. CoinDock does not forecast prices.

Candlestick Charts — Anatomy, Patterns, and What the Evidence Says