Check whether your bullish Crab's fixed 1.618 completion at D has actually confirmed before you trade the pullback.

Quick answer

What is a bullish Crab pattern, and can AI grade it?

Scott Carney formalized the bullish Crab in 2000: a five-point XABCD pattern where D completes at a fixed 1.618 extension of the XA leg, a single ratio rather than a range, making it the deepest completion point in the harmonic family. SnapPChart never calls that turn; it grades the pullback only once a confirmation signal and a new uptrend have already printed above D.

Five points carry the Crab's structure, X, A, B, C, and D, the same labeling Carney used across most of the harmonic family before Darren Oglesby's Cypher and Carney's own Shark changed the formula. The opening decline runs from X to A. B corrects that move somewhere between 38.2% and 61.8% of the XA leg, the identical window a Gartley's B or a Cypher's B can also sit in, so a retracement in that range tells you almost nothing about which pattern you're looking at on its own. C then pulls back 38.2% to 88.6% of the AB swing. Everything that actually separates a Crab from its siblings happens at the very end: D is not allowed to land anywhere across a band the way it does on a Gartley (roughly a 78.6% retracement, still inside the original XA range), a Butterfly (127% to 161.8% beyond X, a wide extension zone), or a Cypher (a 78.6% retracement of the XC leg entirely, a different leg than XA). A Crab's D is pinned to one number: a 1.618 extension of XA, full stop, usually arriving alongside a secondary confluence where CD has also stretched 2.24 to 3.618 times the length of BC. That 1.618 print is the single deepest, most extended completion point of any pattern in this family, further past the starting leg than even Butterfly's widest reading. The part worth sitting with is that a fixed ratio, rather than loosening the read, tightens it: Carney's own writing on the Crab, echoed by multiple independent chart-pattern references, singles it out as one of the more precise potential reversal zones in harmonic trading, precisely because there is one number to hit rather than a wide band to land inside. The deepest swing in the family produces the narrowest target. None of that licenses buying the print the instant price arrives there. A decline that has only just reached the 1.618 extension, with nothing else on the chart, is still a decline in progress by SnapPChart's read, and that restraint mirrors what the Crab's own trading literature already recommends: wait for the market to actually turn off that zone, a reversal candle, a bullish divergence, or a broken trendline, before treating D as anything more than a level worth watching. SnapPChart holds to that same line. Grading switches on only once one of those signals has printed at D and a fresh uptrend, higher highs stacked on higher lows, is already running above it; what gets graded is the first pullback inside that confirmed advance, never the raw tag of the extension zone itself. Upload a screenshot with the X, A, B, C, and D points marked, the Fibonacci ratios labeled at each leg using your own charting platform's tools, the confirmation signal visible, and the new uptrend already in view, and SnapPChart reads that structure back: whether the ratios actually converge the way a Crab's should, whether the confirmation genuinely held, how the volume behind the move looks, and whether the higher highs and higher lows are real. From there it returns a setup grade, an entry on the pullback, a stop placed just beyond the 1.618 extension of XA past D, multiple exit targets, and the case against the trade. It will not plot the five points for you, calculate any of the retracement or extension math, scan a live feed hunting for Crabs, or decide on its own that the reversal at D has happened. That reading, start to finish, stays yours to do before the chart ever reaches an upload button.