Crab pattern — Scott Carney harmonic X-A-B-C-D reversal

The Crab is a harmonic X–A–B–C–D construction with a D reference at 1.618 XA, beyond X. Scott Carney attributes its discovery to 2000. Measure all the selected ratios together: a deep projection does not establish exhaustion, a particularly precise reversal or a trade entry. The examples below separate geometric tests from execution rules.

What the Crab pattern is and where it came from

Carney’s regular Crab uses B from 0.382 to 0.618 XA and C from 0.382 to 0.886 AB. Its XA reference differs from the Gartley’s 0.786 retracement or the Bat’s 0.886 retracement at D. Beyond-X completion is a geometric distinction, not proof that the market has exhausted itself.

The basics of trading harmonic patterns explain the different leg denominators. For a bullish construction, use AB/XA for B, BC/AB for C, AD/XA for D and CD/BC for the final projection. Specify the pivot rule, tolerance and author or tool variant before viewing the outcome.

Structure and Fibonacci levels

The Crab requires its complete set of X-A-B-C-D ratios. A 1.618 completion reference alone does not establish the pattern or a profitable reversal.

The formation has five points joined by four legs. The A-B leg is the first correction after the opening move and defines point B on a shallow retracement, between 0.382 and 0.618 of the XA leg. The B-C leg then retraces between 0.382 and 0.886 of the AB leg.

Volume One names BC projections 2.618, 3.14 and 3.618 for the regular Crab; these named levels are not automatically every value in a continuous interval. It also discusses an equivalent AB=CD minimum and alternate 1.27 or 1.618 AB projections, with less emphasis than in some other patterns. State the selected version and tolerance. A broad zone enclosing different projections is not the same as a common intersection of their tolerances. Use Fibonacci retracements and Fibonacci extensions to check each formula; shared anchors do not make the measurements independent confirmations.

AB = 185.4 pips, BC = 114.6 pips and CD = 414.6 pips. CD/BC ≈ 3.617801, close to the named 3.618 level; that BC projection gives 1.08145772, 0.0228 pips below the XA reference. But CD/AB ≈ 2.236246. A 1.618 AB projection from C gives 1.09292228, 114.6228 pips above D. The candidate therefore does not satisfy a rule requiring that alternate AB projection to coincide within a small tolerance. Several passing ratios do not justify silently ignoring a failed chosen condition.

How to recognise the formation step by step

Step 1 — find the X-A opening leg

Choose ordered, confirmed points on a stated interval and linear scale. Record when each pivot could be known. A visually clear impulse can make an illustration easier to follow, but it does not prove greater predictive reliability or that a choppy market cannot contain the defined geometry. Do not move X or A after seeing the later reaction.

Step 2 — measure the A-B and B-C corrections

Check that the A-B leg ends point B between 0.382 and 0.618 of the XA leg — a relatively shallow correction. Then measure the B-C leg, which should retrace between 0.382 and 0.886 of the AB leg, without letting point C run beyond point A.

Step 3 — confirm point D with a double measurement

Calculate projected levels from the chosen earlier points before considering an entry. Compare each named BC projection and the selected AB=CD version with the 1.618 XA reference, using a tolerance set in advance. The realised turning-point extreme may become identifiable only after later prices; it cannot be assumed to have been an available earlier entry.

Entry, stop and targets — a hypothetical example

To test a trade, define entry confirmation, its availability time, quote sides, stop, exits and costs separately. A projected D is not itself an order or an observed fill. In a separate fictional 10,000-EUR position, assume an actual buy fill at Ask 1.08200, a stop sale fill at Bid 1.08000 and a target sale fill at Bid 1.08600. Gross outcomes are −USD 20 or +USD 40; with USD 3 total opening/closing fees they become −USD 23 or +USD 37. A worse stop fill at 1.07950 produces −USD 28 after those fees. Spread is already represented by the assumed fill sides; do not subtract it again.

Targets at 0.382 and 0.618 of AD measured from D are additional test rules, not inherently conservative exits. In the geometry above AD = 485.4 pips: they project 1.10000228 and 1.11145772, about 185.4228 and 299.9772 pips above D. Actual reward depends on entry and costs as well as the stop and target. In the separate trade model, delaying the buy to 1.08250 while keeping the same assumed sale fills and USD 3 fees changes net outcomes to −USD 28 or +USD 32. Neither geometry nor a tighter stop establishes a reversal probability.

The C anchor affects a BC projection even while the XA reference stays fixed. With X, A and B unchanged, D from 3.618 BC equals C − 3.618(C − B) = 3.618B − 2.618C.

Raising C by 10 pips lowers this BC projection by 26.18 pips. This is algebra under fixed assumptions, not evidence that any row reverses or passes all Crab conditions. Anchor selection and tolerance change the test result; they do not improve it merely by making the displayed zone narrow.

The most common mistakes when trading the Crab

1. Identifying a full pattern from D alone or mixing a Butterfly variant with a regular Crab rule.
2. Assigning a fill at a turning-point extreme that became known only after later prices.
3. Treating every BC coefficient between named projection levels as a match, or overlooking the chosen AB=CD condition.
4. Equating a geometric stop distance with a guaranteed realised loss, without quote sides, size or costs.
5. Moving C, D or the tolerance after seeing the outcome instead of retaining the original candidate and failed checks.

The Deep Crab — a deeper variant

The Deep Crab retains the 1.618 XA reference at D but uses B near 0.886 XA, unlike the regular Crab’s 0.382–0.618 band. Its named BC levels and AB=CD emphasis also differ. A deeper B alone does not establish the Deep Crab variant with its 0.886 B point. Check its remaining conditions; there is no evidence here that its zone is inherently narrower or its reversal more profitable.

Who this pattern is for

Discretionary pivot choices can vary between analysts. A fully specified pivot rule, availability time, formulas and tolerance make the arithmetic reproducible. “Zones rather than points” still needs a numeric tolerance; it does not justify allowing any later price to count. A stop and position size belong to the tested execution plan. An ordinary stop can slip, so a projected level does not set a guaranteed maximum loss.

Compare complete conditions across the Gartley pattern, the Butterfly pattern and broader technical analysis. A rare-looking or deep construction is not evidence of a trading advantage. Keep failures and rejected candidates, separate geometry from the entry/exit rules, and assess unchanged rules on later data.

What to do tomorrow to start learning the Crab pattern

1. Choose a series, interval, linear scale and fixed pivot-confirmation rule. Record when X, A, B and C become known.
2. Calculate B and C ratios, the 1.618 XA reference, each chosen named BC projection and the AB=CD version. Define tolerance before seeing later outcomes.
3. Record passed and failed conditions, projected versus realised D, entry availability, assumed or actual fills, costs and all outcomes.
4. Use an alert only to locate a candidate for review. It is neither proof of a full Crab nor a guarantee of a timely quote or fill.
5. Evaluate frozen rules on later data and retain failures. Twenty demo trades or a high observed hit rate alone does not establish future profitability or readiness for live trading.