Harami: The Rarest Pattern We Measured

A harami is a small candle whose body sits entirely inside the previous large one — the inverse of an engulfing. In our EUR/USD sample it appeared 14 times, the least of any model tested, and the next day closed higher in 50% of cases against a base rate of 44%.

How harami is defined

The current candle's body sits inside the previous body and is materially smaller — we use a threshold of one half. Shadows are ignored, unlike the inside bar where highs and lows are compared. That distinction matters: harami is about bodies, the inside bar about ranges, and one candle can be both or only one of them.

What happens inside the pattern

After a large directional candle the market compresses: the move did not continue, but no retracement occurred either. This is read as fading momentum. In substance a harami is the same pause as an inside bar, measured by bodies rather than ranges, and the conclusions are the same: a stop rather than a reversal.

Entry and stop placement

Harami has practically no standalone entry rules — the model is too weak to build a trade on. Its only sensible use is as a supporting sign: if you were already expecting a move to end at a significant level, a harami adds a little confidence. As the sole basis for entry it does not qualify.

What the measurement showed

The formal definition was run over 400 daily bars of EUR/USD (2025-06-16 to 2026-07-20). Any result has to be compared against the sample's base rate of up days, which is 44%, not against half.

• Harami: 14 occurrences, next day closed higher in 50%

The practical conclusion about harami is not in the percentage but in the rarity. A model that appears a dozen or so times in eighteen months cannot be the basis of a system: even if it worked perfectly there would be too few trades for the edge to show up in an account. That is a separate selection criterion, often forgotten — a pattern must not only work but also occur.

Fourteen cases in four hundred bars makes this the rarest model tested. At that frequency even a correct reading would not have had time to demonstrate itself. The figures describe one sample and one instrument rather than a property of the pattern in general — why that distinction matters is set out in our piece on sample size. The data is open and the method is in how we run backtests.

This material is educational and is not individual investment advice. Trading forex carries the risk of losing capital.