Candlestick formation · Bearish
What is it, in one line
Two red candles with a gap, then a green candle filling it.
Bearish · 3 candles · verdict: rule and thresholds below
Two red candles with a gap, then a green candle filling it. The engine checks the formation as an executable condition over open, high, low and close, with a trend filter applied before the count is kept.
Method
The mirror: two down candles with a gap between them, then an up candle opening inside the second and closing inside the first.
The traditional claim
Traditionally read as a bearish continuation.
How this detection is wrong
See upside-gap-three-methods — the gap is filled here and survives in the tasuki version.
Published so the count can be reproduced. Every figure below is a parameter of the rule, not a preference: change one and the occurrence count changes with it.
| Parameter | Meaning | Default |
|---|---|---|
| trendPeriod | bars of prior trend inspected | 5 |
| maxPriorSlope | prior slope must be below | -1 |
| minGapAtr | gap between bars, at least, in ATR | 0 |
| minBodyAtr | body, at least, in ATR | 0.3 |
| atrPeriod | bars of ATR used to size everything below | 14 |
These share enough geometry that a detector can return more than one of them on the same bar. The engine reports every match rather than picking a winner.
Every pair on the site that can be mistaken for another, and what separates each: the look-alikes index.
The mirror: two down candles with a gap between them, then an up candle opening inside the second and closing inside the first. The detector also requires a prior downtrend, so the same shape in the wrong context is not counted.
It is read as bearish. Traditionally read as a bearish continuation.
See upside-gap-three-methods — the gap is filled here and survives in the tasuki version.
As an executable rule over open, high, low and close, with 5 named parameters — including bars of prior trend inspected 5, prior slope must be below -1, gap between bars, at least, in ATR 0 — all published on this page so the count can be reproduced. Strict and loose variants are defined as well, so the sensitivity of the count to those thresholds is visible rather than assumed.
Measurement pending re-run
No hit rate is published here yet. The run that would supply one was computed on engine ca6d9e36c81c; the engine is now cfb08765de7c. Its candlestick rows included the still-forming bar, and it predates the chart formations entirely, so those figures have been withdrawn rather than published with a caveat. The definition above is unaffected: it is what the detector tests, and no re-run changes it.