Free lesson · Technical analysis

Harmonic patterns: ratio constraints and competing candidates

Open interactive lessonPractice calculationsExplore labs

• Start with an ordered X–A–B–C–D pivot sequence. Each pivot must have a recorded confirmation time.
• Gartley, Bat, Butterfly and Crab templates use different combinations of retracements and extensions. Publish the exact template table and tolerances used.
• For a common Gartley convention, B is near .618 of XA and D near .786 of XA; other leg constraints must also pass. Two ratios alone are not a full detector.
• A Bat convention often uses a shallower B retracement and D near .886 of XA; Butterfly/Crab conventions extend D beyond X. Definitions vary across implementations.
• Check pivot order, alternating direction, leg length, time spacing and every required ratio. Group near-duplicate candidates from overlapping pivot choices.
• Use tolerance bands fixed on validation. Compare a ratio-constrained detector against the same swing geometry without the Fibonacci restrictions.
• A complete pattern at D is observable only when the D pivot confirmation rule passes. Historical plotting at the D extremum does not mean trading there was possible.

Use the rule

• Fix the definition, units and information timestamp.
• Compute the example and inspect the graph.
• Compare with a simple baseline on untouched periods after costs.

Before moving on

Write the exact rule, its availability time, an invalidation condition and a fair out-of-sample test.

• Research context: review the related evidence checkpoint. The numerical convention here defines a candidate feature; that related research does not validate this exact rule.

def harmonic_leg(x,a,b,target=.618,tolerance=.03):
    if a==x or tolerance<0: raise ValueError("Nonzero anchor leg and nonnegative tolerance required")
    ratio=abs(b-a)/abs(a-x)
    return ratio,abs(ratio-target)<=tolerance

print(harmonic_leg(100,120,107.64)) # one constraint, not a full Gartley detector