Candlestick patterns: what statistics say

[Shadow Trader]

Hammers, engulfing candles and dojis fill trading books. The claims are bold: this pattern signals a reversal with high probability.

Large-scale studies on currency data are less enthusiastic. Many patterns show forecasting power close to random once costs are included. A few, in specific contexts, produce a small edge. Context seems to matter far more than the shape: a pin bar at a major support level in a range behaves differently from a pin bar in the middle of a trend.

Another issue is subjectivity. Two traders looking at the same candle can disagree on whether it qualifies, which makes testing hard.

The pragmatic view treats candlesticks as a compact description of what price did during an interval, showing rejection or acceptance of a level. They are a language, not a prediction engine.

Rules that define a pattern in numerical terms, such as wick at least twice the body, make testing possible.

Has any candlestick pattern been tested here with objective rules, and did it hold up?