Elliott Wave Radar

Features

• ZigZag pivot engine with ATR noise filter, two degrees (primary + sub)
• Hard-rule validation:
  • W2 never retraces beyond the start of W1
  • W3 is never the shortest of W1/W3/W5
  • W4 never enters W1 price territory (toggle for diagonals)
  • W3 exceeds the end of W1, W5 exceeds the end of W3 (no truncation in v1)
• Fibonacci guideline scoring (0–100): W2 ≈ 0.5–0.618, W3 ≈ 1.618×W1, W4 ≈ 0.382×W3, W5 ≈ 0.618–1.0×W1, plus an alternation bonus (W2 vs. W4 depth)
• Corrective patterns:
  • Zigzag A–B–C (B retraces 0.236–0.886 of A, C travels beyond A)
  • Flat (regular/expanded: B retraces 0.90–1.38 of A) — disjoint thresholds, so zigzag and flat never compete on the same structure
• Developing counts with Fib target projections:
  • W3 setup after W1–W2 confirm (targets ×1.0 / ×1.618 / ×2.618 of W1)
  • W5 setup after W1–W4 confirm (targets ×0.618 / ×1.0 / ×1.618 of W1)
  • C-wave setup after a completed impulse + confirmed A–B (targets ×1.0 / ×1.618 of A; invalidated if price takes out the impulse extreme)
  • Post-impulse correction targets (38.2 / 50 / 61.8 % of the impulse)
• Invalidation level drawn and monitored bar-by-bar — a broken level kills the count and fires an alert
• Sub-wave labels (i)–(iv) / (a)–(b) from the finer pivot degree; the dashboard reports whether W1/W3 internally divide into 5 legs
• Alternate (runner-up) interpretation shown in the dashboard
• Alerts: impulse complete, ABC complete, W3 setup, W5 setup, count invalidated

How counting works

1. Primary pivots come from ta.pivothigh/pivotlow (confirmation delay = pivot length) with an ATR minimum-swing filter against noise. Each accepted pivot is then re-anchored to the true extreme since the previous pivot — sharp counter-swings too short to confirm as pivots still mark the correct wave point (e.g. the strongest Wave-2 pullback, not a later weaker high).
2. On every new confirmed pivot, six interpretations ending at that pivot are evaluated: complete impulse (6 pivots), complete zigzag A–B–C (4), complete flat (4), post-impulse C-wave setup (3), developing W5 setup (5), developing W3 setup (3).
3. Each interpretation must pass its hard rules, then gets a Fibonacci guideline score. Candidates below the score gate (input, default 55) are discarded.
4. The best candidate wins (complete patterns get a completeness bonus over developing prefixes); the runner-up appears as “Alt count” in the dashboard.

Honesty constraints (by design)

• Counts use confirmed pivots only — labels appear Pivot Length bars after the actual extreme. No repainting, but also no real-time pivot calls.
• Only the most recent structure is labeled. Historical counts are not preserved when a newer structure replaces them.
• One wave degree is counted; the sub-degree is used for verification and sub-labels, not for an independent second count.
• Truncated fifths, triangles and complex corrections (W-X-Y) are not modeled — such phases simply show “No valid count”. Flats are supported since v1.1.

Inputs

Group	Input	Default	Purpose
ZigZag Engine	Pivot Length (primary)	8	Swing size of the counted degree
ZigZag Engine	Pivot Length (sub)	3	Finer degree for sub-wave checks
ZigZag Engine	Min Swing Size (ATR ×)	0.5	Noise filter for primary pivots
Wave Validation	Allow W4 into W1 (diagonal)	off	Relaxes the overlap rule
Wave Validation	Min count score	55	Quality gate — below this, no count is shown
Display	Projections / sub-waves / dashboard / colors	on	Visual toggles