Shark pattern — Carney's O-X-A-B-C harmonic reversal

Shark is Scott Carney’s harmonic-pattern specification, first released in 2011 according to the author. Its O–X–A–B–C labels and ratio anchors must be explicit: writing “0.886” beside a chart does not identify the measured segment. Separate geometry, the time each point became known and the rules of a proposed trade. Matching ratios does not establish profitable performance.

What the Shark pattern is and where it came from

Carney uses 0 for the starting point; here O denotes that same point, followed by X, A, B and C. Do not shift these names into X–A–B–C–D. Leg lengths are absolute price differences, separate from elapsed bars. The bullish label describes a proposed reaction after C, not a rule that the preceding BC leg rises.

The basics of trading harmonic patterns explain segment ratios. For this pattern, record the source, each numerator and denominator, and any tolerance before inspecting outcomes. Detectors can implement different specifications; a matching name alone does not make their results interchangeable.

The O-X-A-B-C structure and Fibonacci levels

We use the bullish variant in Harmonic Trading: Volume Three, printed pages 119–120: XA/OX is 0.382–0.618, AB/XA is 1.13–1.618, BC/AB is 1.618–2.24, and BC/OB is 0.886–1.13, measured from B toward O. OX rises, XA falls, AB rises beyond X and BC falls toward or beyond O. A bearish construction reverses the leg directions and requires its distances to be recalculated.

The book’s bullish PRZ diagram labels its final band against 0B. Harmonic Pattern Collection instead describes C relative to OX. These are different measurement specifications. The Fibonacci retracements tool still needs an explicit start, end and projection direction; do not combine both descriptions into one unexplained rule.

How to recognise the formation step by step

Step 1 — find the O-X leg and the context

For a separate qualifying example, assume arbitrary prices O = 100, X = 120, A = 110, B = 125 and C = 98. Lengths OX, XA, AB, BC and OB are 20, 10, 15, 27 and 25. The four ratios are 0.50, 1.50, 1.80 and 1.08, each within our selected bands. XC/OX would instead be 22/20 = 1.10. That different number happens to pass, but it does not make the anchors interchangeable. These are fictional units, not EUR/USD quotes.

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

Once B = 125 is known, the OB condition gives C = 125 − r × 25 for r from 0.886 to 1.13: [96.75,102.85]. The AB condition gives C = 125 − q × 15 for q from 1.618 to 2.24: [91.40,100.73]. Their intersection is [96.75,100.73]. C = 98 passes both; C = 102 passes BC/OB = 23/25 = 0.92 but fails BC/AB = 23/15 ≈ 1.5333. One passing ratio does not rescue a failed condition. No tolerance or time-to-arrival rule is included.

Step 3 — retain the measurement and confirmation times

Changing only C to 97 gives BC/OB = 28/25 = 1.12, within the band, but XC/OX = 23/20 = 1.15, outside it. BC/AB = 28/15 ≈ 1.8667 still passes. The two final ratios are equal only at C = 100 in this model. Separately, suppose our pivot rule requires two lower highs on each side of a peak or two higher lows on each side of a trough. We assume those local bars satisfy the rule: B on bar 22 is confirmed at the close of bar 24; C on bar 26 at the close of bar 28. Plotting C back on bar 26 cannot move confirmation information there. This is a chronology example, not a complete OHLC dataset.

Entry, stop and targets — a hypothetical example

Let us return to the setup in the table above. At C = 1.0865, the example is not a valid Shark under these requirements: BC is 235 pips and AB is 150 pips, so BC/AB ≈ 1.567, below the required 1.618 minimum. Although BC/OB = 235/250 = 0.94 passes the other band, one valid ratio does not rescue the failed condition. Do not label this candidate a completed Shark or use its geometry alone as an entry instruction.

In the separate qualifying 100–120–110–125–98 model, halfway back through BC is 98 + 27/2 = 111.50. This is an arithmetic reference, not an assured target or return. An entry at first band contact and an entry after pivot confirmation require different decision times and executable prices. Specify exit, spread, commission, funding and ordinary-stop execution assumptions before testing. A later confirmed low of 98 cannot simply be assigned as the fill of the confirmation strategy.

A scanner also needs a reproducible pivot-selection rule, bar interval, data source and tolerance. Retain all recognized and rejected candidates, including those without a trade or reaction. Moving O or A after a failed outcome changes the specification; it does not repair the original test.

The most common mistakes when trading the Shark

1. Confusing labels or anchors. Write each ratio as numerator/denominator and retain the chosen source; OX and OB are not interchangeable.
2. Accepting one passing condition. Our C = 102 passes BC/OB but fails BC/AB, so it is rejected under the full selected specification.
3. Ignoring the symbol grid. The preserved diagram’s positive-width intersection contains no permitted price under its assumed 0.001 step. Rounding outside the interval does not preserve the conditions.
4. Treating an ordinary stop as a maximum-loss promise. Record the trigger, buffer, costs and worse-fill assumptions separately from geometric invalidation.
5. Selecting the exit after seeing a reaction. A time limit, partial exit and price target are different rules; none follows as universally profitable from the name Shark.

How the Shark links to the 5-0 pattern

The 5-0 pattern has its own structure and later retracement measurement. Carney’s relationship between the patterns is a definition to examine, not a measured probability that a Shark becomes 5-0. Record the later observation separately; do not use it to choose earlier Shark points.

The Crab pattern uses a different X–A–B–C–D specification with its XA extension at D. The shared number 1.618 is insufficient to identify either pattern: names, measured legs and all other conditions matter.

Who this formation is for

The tools of technical analysis can help define a research procedure, but adding momentum or horizontal levels does not by itself establish an edge. Compare a fixed rule after costs on data not used to choose its parameters. A set number of successful demo trades cannot certify live readiness or remove execution risk.

What to do tomorrow to learn the Shark pattern

1. Record the specification and information times. Save source, ratios, pivot-selection method and the bar when each point became confirmed.
2. Recalculate every ratio and the interval intersection. Then check whether the actual symbol grid contains a permitted price; preserve rejection reasons.
3. Record all candidates, including no-trade cases. Keep decision time, actual or estimated fill, costs, exits and rule version; demo results remain simulated.
4. Define the decision and exit before evaluating results. An alert only signals a condition; it does not confirm a pivot, provide a fill or establish a profitable reversal.
5. Separate parameter development from evaluation on unused data. Include failed patterns, costs and worse execution; do not choose an arbitrary twenty-trade threshold as proof of an edge.