What is the difference between Crab and Deep Crab?

The main distinction is B: the regular Crab uses a 0.382 to 0.618 XA retracement, while Deep Crab uses about 0.886 without crossing X. Both retain the 1.618 XA completion reference. Their BC and AB=CD checks also need to be evaluated under the chosen definition.

Is the 1.618 Crab extension measured from X or A?

Measure AD against XA. With bullish X=100 and A=200, D=200 - 1.618 × 100=38.2. This places D 0.618 XA beyond X. Subtracting 1.618 XA from X uses the wrong origin.

What tolerance should a Crab detector use?

There is no tolerance specified here as universally correct. CRAB-1 uses an explicit 0.010 absolute allowance around AD/XA=1.618 and 0.020 around selected BC ratios. These are teaching settings. Record the chosen bands and keep them fixed when evaluating candidates.

Does a price touch in the PRZ confirm a reversal?

No. A PRZ touch, a confirmed pivot and a trading trigger are different events. A pivot requiring two right-hand bars is unavailable until those bars complete. Even a candidate passing all chosen rules can fail after entry; this page makes no win-rate claim.

TradingView chart patterns

Crab harmonic pattern: check Crab and Deep Crab ratios

The Crab harmonic pattern is a five-point XABCD reversal candidate defined by D extending to 1.618 of XA, measured from A. Scott Carney distinguishes the regular Crab, with B retracing 0.382 to 0.618 of XA, from the Deep Crab, whose B reaches about 0.886. Both finish beyond X. Converging projections identify a potential reversal zone (PRZ); they do not establish that price will reverse.

See the decision guide

Crab versus Deep Crab: the original reference ratios

This table uses the named BC projection levels in Carney's current Crab and Deep Crab webpages, with the C range from Harmonic Trading: Volume One. The author gives AB=CD less weight in the regular Crab and more weight in the Deep Crab. These references describe geometry, not a guaranteed reversal.

Source variations matter: Volume One also describes Deep Crab BC projections from 2.0 and notes occasional 2.0 or 2.24 projections in regular Crabs. The discrete levels below are a narrower selection; do not treat them as the only convention the author has published.
A bullish candidate has time-ordered X, A, B, C, D with D < X < B < C < A. Reverse the price inequalities for a bearish candidate. Reject equal-price or zero-length denominator legs.
Use absolute price differences for all ratios. AD/XA means |A - D| / |A - X|, not |X - D| / |A - X|.
Source conflict: TradingView's XABCD drawing help lists Crab BC extensions of 2.24 or 9.64. Those values are not the regular-Crab set in the Carney excerpt used here. Use that help page for drawing controls; do not merge its ratio list into this worksheet.

Measurement	Regular Crab reference	Deep Crab reference
AB / XA	B retracement: 0.382 to 0.618	B retracement: 0.886; may extend deeper without crossing X
BC / AB	C retracement: 0.382 to 0.886	C retracement: 0.382 to 0.886
AD / XA	D extension: 1.618	D extension: 1.618
CD / BC	2.618, 3.14 or 3.618	2.24, 2.618, 3.14 or 3.618
CD / AB	At least AB=CD; 1.27 and 1.618 alternate projections are secondary checks	AB=CD or an alternate projection; the description emphasizes 1.27

Make the tolerance a declared rule

CRAB-1 is an illustrative geometry screen for the regular Crab. Its bounds below are teaching choices, not a universal tolerance, a TradingView default or a complete implementation of Carney's trading method. Test every row using unrounded prices and one consistent wick or close convention.

The D allowance is an absolute ratio error of 0.010. With XA=100, it permits one price unit on either side of the D reference. It is not a 1% relative error around 1.618.
The BC test creates three separate bands: 2.598 to 2.638, 3.120 to 3.160, and 3.598 to 3.638. A value of 2.90 fails even though it lies between the lowest and highest reference numbers.
A Deep Crab detector needs a separately named B rule. For example, 0.886 ≤ AB/XA ≤ 0.896 is an explicit teaching choice that allows a slightly deeper B while keeping it inside XA. Do not silently apply the regular Crab band to it.
Passing CRAB-1 does not establish close agreement with an alternate AB=CD target. The example below exposes that difference instead of hiding it in a single pass label.

Test	CRAB-1 inclusive acceptance rule
B retracement	0.382 ≤ AB/XA ≤ 0.618
C retracement	0.382 ≤ BC/AB ≤ 0.886
D extension	\|AD/XA - 1.618\| ≤ 0.010
BC projection	Within 0.020 of one of 2.618, 3.14 or 3.618
Minimum CD completion	CD/AB ≥ 1; report the 1.27 and 1.618 projection prices separately

Recalculate a bullish Crab candidate

Use constructed prices X=100 at bar 0, A=200 at bar 10, B=138.2 at bar 20, C=176.4 at bar 30 and D=38.2 at bar 40. These are not market observations. They produce XA=100, AB=61.8, BC=38.2, CD=138.2 and AD=161.8. The candles between the points are unspecified, so these coordinates test geometry only.

Project from A: 200 - 1.618 × 100 = 38.2. D is 61.8 below X, so XD/XA is 0.618 while AD/XA is 1.618. Starting the subtraction at X would give -61.8 and the wrong completion level.
The bearish mirror uses 300 minus each price: X=200, A=100, B=161.8, C=123.6 and D=261.8. All absolute ratios are unchanged; D lies above X.
Replacing B with 111.4 gives AB/XA=0.886 but leaves BC/AB=65/88.6=0.733634 and CD/BC=138.2/65=2.126154. It fails every named Deep Crab BC band in this table with a 0.020 allowance. Under that discrete-band rule, moving B alone is insufficient; a broader convention needs its own checks.

Measurement	Calculation	Result
B	61.8 / 100	0.618: passes at the upper bound
C	38.2 / 61.8	0.618123: passes
D	161.8 / 100	1.618: passes
BC projection	138.2 / 38.2	3.617801: within 0.020 of 3.618
CD / AB	138.2 / 61.8	2.236246: exceeds minimum; not near 1.27 or 1.618

Build the PRZ from prices, not a pattern label

Before D exists, the fixed XABC points let you project possible completion prices. For this bullish example, subtract the appropriate leg multiple from A or C. Compare the levels rather than declaring that every Fibonacci calculation agrees.

The XA and selected BC reference prices differ by 0.0076. Their narrow envelope is a two-measurement PRZ. The alternate AB=CD levels are far above it; this is not three-way confluence or an ideal match to every published relationship.
CRAB-1 gives an XA price band of 37.2 to 39.2. Its selected BC band is 37.4284 to 38.9564, because 0.020 × 38.2 = 0.764. Their intersection is 37.4284 to 38.9564. Other geometry tests still apply.
If D becomes 39.1 with XABC unchanged, it passes the XA band but CD/BC becomes 137.3/38.2=3.594241 and fails the selected BC band. The broader XA allowance cannot override that failure.
For a detector using both bands, crossing below the intersection while seeking a bullish completion rejects that candidate. A future candidate must be recorded separately rather than moving old anchors to fit the outcome.

Projection	Calculation	Projected price
XA reference	200 - 1.618 × 100	38.2000
Selected BC reference	176.4 - 3.618 × 38.2	38.1924
Alternate 1.618 AB=CD	176.4 - 1.618 × 61.8	76.4076
Alternate 1.27 AB=CD	176.4 - 1.27 × 61.8	97.9140

A historical D label is not a real-time signal

Assume X through C are already confirmed. For a separate timing exercise, define a low pivot as strictly lower than the lows of two completed bars on each side, rejecting ties. D at bar 40 cannot qualify until bar 42 closes. The coordinate example contains no surrounding candles and therefore does not itself establish this confirmation.

For an optional trigger exercise, assume the D-candle high is 42. Require a close above 42 on bars 43 through 47. Invalidate first if any later bar touches 38.2 or lower, even if that same bar closes above 42. Expire the candidate after bar 47 without a trigger.
These trigger, retest and expiry rules are illustrative additions, not the harmonic definition. Reverse their inequalities for the bearish mirror.
A label drawn back on bar 40 must retain bar 42 as its first confirmation time. Backtests that enter at the historical pivot before the right-hand bars exist use information unavailable at the claimed entry.
A geometry failure or trigger is not an order fill or a guaranteed loss limit. A strategy still needs position sizing, order timing, exits, costs and slippage assumptions. Keep failed candidates in any evaluation.

Event	What is known at that time
Bar 40 reaches 38.2	The projected zone has been touched; D is provisional
Bar 41 or 42 has a low of 38.2 or less	The proposed strict pivot at bar 40 fails
Bar 42 closes with all four neighboring lows above 38.2	D can be confirmed if the geometry also passes
Bar 43 or later closes above the D-candle high	A separately chosen price trigger may occur

Keep Crab separate from Bat, Gartley and Butterfly

Bat and Gartley are internal retracement structures, while Crab completes beyond X. Butterfly is also an extension structure, so an outside D alone cannot distinguish it from Crab. Compare the selected B and D definitions together and recalculate the remaining legs; do not rename a failed pattern after seeing its eventual reversal.

The Deep Crab B near 0.886 should not be confused with the Bat D reference near 0.886. The point name and denominator determine what the number means.
In the fixed XA=100 to 200 example, the Crab D reference is 38.2 for both variants. Deep Crab changes the B reference to 111.4; it does not move D to a 0.886 retracement.

Review the worksheet in TradingView and Pineify

TradingView's XABCD drawing tool accepts five manually placed anchors and allows price and bar coordinates to be edited. Place the measured points, then save the symbol, timeframe, price convention and detection timestamps with the worksheet. A manually chosen drawing does not prove that a scanner would have found it in real time.

Use AI Chart Analysis to review an annotated screenshot with readable prices and timeframe. Verify exact ratios from the worksheet; visual analysis can misread anchors and does not certify a Crab or Deep Crab.
Give Pine Script AI Coding Agent the ratio bands, pivot rule, alternate-projection treatment and trigger order to draft a detector. Ask it to report both the pivot bar and the first confirmation bar.
Inspect and syntax-check generated code, then compare its outputs with hand-calculated passing and failing cases. Compilation, a plausible chart label and an AI response do not demonstrate trading performance.

Where Pineify fits

AI Chart Analysis

Review an annotated XABCD screenshot, then check the interpretation against your measured prices and confirmation times.

Also useful: Pine Script AI Coding Agent. Draft a detector from your explicit ratio bands and pivot rules, then inspect and test its output.

Review an annotated chart

Related resources

TradingView chart patterns Bat internal D retracement Gartley ratio checks Butterfly extension rules

Frequently asked questions

Educational information, not investment advice. CRAB-1, all price coordinates and the confirmation exercise are hypothetical. They are not a backtest or a complete trading method. A ratio match, pivot confirmation or AI review does not guarantee a reversal, fill price or profit.

Sources and verification

Last verified: 2026-10-17

Published by Pineify. Pineify is an independent product and is not affiliated with TradingView. Sources support the technical explanations; examples do not establish trading performance.