Harmonic pattern calculator: ratios and tolerance

Harmonic pattern ratios used here

These bands make the calculation reproducible. They are a partial geometric screen, not every condition in the source methods. The table shows base ratios before tolerance; intervals accept all values between their endpoints.

Gartley and Bat complete inside XA. Butterfly, Crab and Deep Crab complete beyond X. This tool uses the 1.27 XA Butterfly convention and the 2.618 to 3.618 BC band for Crab. Bat uses the narrower 0.382 to 0.500 B band. Other implementations may choose different bands, so compare definitions before comparing signals. Deep Crab's 2.0 lower BC bound follows the Deep Crab chapter in Harmonic Trading Volume One.

Cypher requires C beyond A and uses XC/XA for its extension, then CD/XC for the 0.786 retracement. Shark also extends C beyond A in this interface. Its native labels 0, X, A, B, C map to our X, A, B, C, D. No fixed B retracement band is imposed for Shark, but B must remain between X and A. The Shark impulse structure in Harmonic Trading Volume 3 uses the 1.13 to 1.618 and 1.618 to 2.24 extensions. Shark and 5-0 have different completion rules.

Choose anchors before checking ratios

1. Fix the symbol, session, timeframe and price scale. This calculator measures raw price differences, not logarithmic distances.
2. Select successive pivots in time order, using the same high/low or close-based rule throughout. For a bullish candidate, XA rises, AB falls and BC rises; reverse the directions for a bearish candidate.
3. Keep B strictly inside XA. C stays between B and A for the five conventional structures; Shark and Cypher require C beyond A. Equal adjacent prices cannot define a leg.
4. Set tolerance before assessing the outcome. Widening it admits more candidates and broadens the projected price bands. It does not improve the evidence for a reversal.

TradingView's XABCD drawing tool places five anchors manually. Use the ABCD pattern guide to distinguish a four-point measured move from a five-point harmonic structure.

Worked example: a candidate is not a reversal

Illustrative prices, not a historical trade: X = 100, A = 200, B = 138.2, C = 169.1. XA = 100, AB = 61.8 and BC = 30.9. That gives AB/XA = 0.618 and BC/AB = 0.500, within the Gartley bands.

The XA projection is D = A − 0.786 × (A − X) = 121.4. Subtracting from A is essential: a retracement starts at the end of XA. The BC projection runs from 169.1 − 1.618 × 30.9 = 119.1038 to 169.1 − 1.27 × 30.9 = 129.8570. At 0% tolerance, the two projections overlap at 121.4.

At 5% tolerance, the XA band becomes 117.4700 to 125.3300. The BC band becomes 116.6040 to 131.8192, so the overlap is 117.4700 to 125.3300. Neither interval says that price will reach it or reverse there. This example also leaves AB=CD symmetry unchecked, so it is not a fully validated Gartley.

Completion, confirmation and invalidation

Completion means that an observed D satisfies the selected ratio and geometry rules. Confirmation is a separate price event. An illustrative bullish rule could require a later bar to close above the D bar's high; the bearish version would require a close below its low. Choose that rule in advance and test its delay, rather than treating a touch of the projected zone as confirmation.

A Gartley or Bat that continues beyond X violates its retracement geometry. Butterfly and Crab already project beyond X, so the same invalidation rule does not apply. Define the allowed final-leg ratio and maximum risk before evaluating an entry. A projection with no overlap, or a final leg outside the chosen limits, is a failed candidate under those rules.

Automated pivots have another delay. A pivot requiring three right-hand bars is only known three bars after its apparent location. Drawing the label back on the pivot does not make it available earlier. TradingView explains this in its repainting documentation. Record the detection bar when testing alerts, use confirmed data, and include costs and failed candidates in any strategy evaluation.

Educational calculations only, not investment advice. No pattern ranking, reversal probability or return estimate is established by these ratios.

Frequently asked questions

Sources and calculation scope

The table links each pattern to its source. TradingView's Cypher documentation supplies the XC-based measurements; Scott Carney's pattern pages and books supply the other comparison conventions. Source definitions are not identical across all publications, and this calculator does not reproduce a complete trading method. Sources checked October 16, 2026.

For individual projection levels, use the Fibonacci extension calculator. For other formation rules, return to TradingView chart patterns.