What Harami is (and what it is not)
Harami is a candlestick pattern made of two candles. The usual description is that the first candle’s body is followed by a second candle whose body is smaller and sits within the body range of the first candle. “Harami” is often presented as a potential reversal clue, meaning traders look for a transition from one short-term direction toward another.
A key limitation starts here: Harami is a visual description, not a built-in rule that forces a specific market response. Whether the market interprets the two candles as meaningful often depends on context (for example, whether the candles occur after a directional move), and on how strictly the pattern is defined by the person or tool that identifies it.
How Harami works in practice
Harami identification usually requires assumptions that can vary:
- Candle-body rules: Some definitions focus on candle bodies rather than wicks, and require the second body to be fully inside the first body range.
- Sequence timing: The pattern is defined across two specific bars; different chart timeframes or candle construction can change what you see.
- Context filters: Many chart readers only treat it as “Harami” when it appears after a move with a clear prior direction.
Because these assumptions are not universal, two observers can look at the same price chart and disagree on whether Harami is present, or on whether it signals anything beyond coincidence.
Evidence and example: why matching the pattern is not the same as predicting a turn
Imagine you mark a Harami after a prior upswing. Even if price later pauses or pulls back, that does not prove the pattern caused the change; it may reflect general market fluctuation.
A practical way to think about this limitation is to separate “pattern presence” from “outcome.” The pattern only describes the relationship between two candles. The subsequent path of price depends on many other variables that are not encoded in the two-candle structure, such as order flow, broader market conditions, and execution effects (including trading costs and how orders fill).
Limitations and failure modes
The main failure modes of Harami come from ambiguity, context dependence, and uncertainty in outcomes:
-
Ambiguous boundaries If your rule for “inside the first candle’s body” differs (for example, whether partial overlap counts), the pattern frequency changes. Different strictness can make the pattern look more or less common without changing the underlying market.
-
Context dependence Harami is often treated as more relevant when it appears after a directional move and near areas where traders expect reactions. If it appears in a range, during low follow-through, or without a clear prior push, the same two-candle structure may be less informative.
-
False positives from random variation Two-candle overlaps can occur in normal price noise. Without a consistent context definition and without measurable filtering, Harami can mark many events that do not lead to the type of turn traders expect.
-
Sensitivity to timeframe and chart construction Because Harami is defined on candles, changing the timeframe can alter which candles form the pattern. The same market episode may produce a clear two-candle Harami on one timeframe and not on another.
-
No guaranteed relationship to future movement Even when a Harami is correctly identified, future results are uncertain. Historical appearances do not establish that the pattern will behave the same way in the future, especially across different volatility regimes.
Verification: what you can independently check
To verify Harami’s usefulness without assuming predictive certainty, you can:
- State your definition: Specify exactly which parts of candles count (bodies vs wicks) and how much overlap is required.
- Specify your context rule: For example, define what counts as a meaningful prior move for your purposes.
- Test consistently on the same timeframe: Measure what happens after Harami instances under the same identification rules.
- Track uncertainty: Include variability in outcomes rather than expecting a uniform result.
This approach addresses the limitations directly: it forces you to separate the stable mechanics of the pattern definition from the variable conditions that determine what happens next.