Triangle Compression Scanner Pro

Shares its core approach with Wolfe Wave Scanner Pro (swing engine, best-of-window search, weighted scoring, multi-pattern tracking, debug/dashboard conventions) but is a separate indicator — triangle geometry (variable-length boundary lines from a swing window) doesn’t fit the same fixed 5-point model a Wolfe wave needs.

Features

• Alternating swing engine (ta.pivothigh/ta.pivotlow) with bounded history
• Sliding-window search: for every possible window end within Suchbereich letzte Swings, the Swings pro Dreieck-Fenster most recent swings up to that point define the two boundary lines — not a fixed point count, so the window can span anywhere from a tight 6-swing triangle to a wide 20-swing one
• Tracks multiple simultaneously valid, non-overlapping triangles at once, each with its own drawings and breakout state. An overlapping candidate only replaces an already-tracked triangle if it scores higher
• A new candidate only starts tracking if price is still between its own boundary lines (plus a small Breakout-Puffer ATR buffer) at the moment it’s detected — otherwise it’s skipped as already broken (pivot-confirmation lag means a lot of candidates are already resolved by the time they’re found)
• Weighted 0-100 score across eight continuous criteria: swing path, traversal, realized compression, boundary proximity, line quality, boundary/apex, duration, relevance — plus a set of hard gates that must all pass before a candidate can fire
• Live breakout tracking: either boundary being closed beyond (plus Breakout-Puffer ATR) resolves the pattern (removes it from tracking) and fires a directional alert
• Score-breakdown debug table for tuning thresholds

Scoring

The upper/lower boundary slopes must first match one of the three valid triangle shapes — symmetric (upper falling, lower rising), ascending (upper flat, lower rising), descending (upper falling, lower flat); “flat” is ATR-normalized against the window’s own width, not an absolute slope value (structureOk). Everything below is then scored continuously and gated separately.

Criterion	Points	What it checks
Swing path	20	Swings must alternate high/low without repeats, and same-side highs/lows must monotonically progress in the triangle’s direction (falling highs for symmetric/descending, rising lows for symmetric/ascending). Gate pathOk additionally requires ≥6 path points, ≥3 highs and ≥3 lows, and a perfect progression match — a single non-progressing pair fails the gate even if the score stays high
Traversal	20	Consecutive swing-to-swing legs must actually cross the local channel width, not just sit inside a generous outer envelope. Score blends the average traversal ratio with the fraction of legs clearing Min. Traversal pro Leg; gate traversalOk requires a run of ≥3 consecutive qualifying legs
Realized compression	15	Compares actual price-leg amplitudes, not just the fitted geometry: the last two legs must be tighter than the first two (≤Max. realisierte Leg-Kompression), and same-direction legs must generally shrink leg-over-leg (≥60% of tested pairs). Gate realizedCompressionOk additionally requires the projected end/start width ratio to clear Ziel proj. Endbreite / Startbreite and at least 5 legs
Boundary proximity	15	Average proximity quality of every non-anchor high/low to its boundary line, in ATR terms (full credit ≤0.25 ATR, zero credit ≥0.75 ATR) — rewards boundaries the whole window actually respects, not just the two points that define them
Line quality	10	Blends boundary residual quality, bar-containment ratio, violation ratio, and contact-zone distribution — draws from the same soft ≥90% bar-containment check as barsInsideOk (no longer a standalone veto, see Breakout below). Gate lineQualityOk: for symmetric triangles, the two boundary slopes must stay within a 3x magnitude ratio of each other; ascending/descending triangles just need structureOk
Boundary / apex	10	60% anchor-span quality (how much of the window each boundary’s own touch points span) + 40% apex-maturity quality (how close the two lines’ projected intersection lands to 65% of the way through the pattern’s own duration). Gates boundaryOk (anchor span ≥Min. Anker-Span / Pattern-Dauer, mean residual ≤Max. mittleres Boundary-Residuum ATR, zero swing violations, contacts spread across the window’s early/middle/late thirds, and — for ascending/descending — an extra touch confirming the flat side) and apexOk (the intersection must lie ahead of the window, within Max. Apex-Distanz× its width)
Duration	5	Continuous credit up to 6× Swing-Stärke bars; gate durationOk still requires at least 3×
Relevance	5	Decays linearly with how many bars have passed since the window’s own end, relative to Max. Pattern-Alter× the window width — an old-but-still-technically-valid candidate scores lower than a fresh one

A pattern only fires when every hard gate — structureOk, pathOk, traversalOk, realizedCompressionOk, boundaryOk, apexOk, lineQualityOk, durationOk — passes and the weighted score clears Mindestqualität.

Breakout

Once a triangle fires, both boundary lines (extended forward) are checked every bar. A close beyond either one by more than Breakout-Puffer ATR resolves the pattern — it’s removed from tracking, a breakout label is drawn, and the corresponding TCS · BULL BREAKOUT / TCS · BEAR BREAKDOWN alert fires. There’s no separate invalidation level like Wolfe Wave Scanner’s point-5 break: for a triangle, a confirmed break of either boundary is the resolution, in whichever direction it happens. The pattern’s own creation bar has a one-bar grace period (no breakout check) so an already-broken pattern doesn’t get labeled as breaking out in the same instant it’s drawn — every bar after that, the same buffered containment check applies, so a pattern that’s outside its lines (beyond the buffer) is always resolved by the next bar.

Multiple patterns

Max. gleichzeitige Patterns (default 5) caps how many valid triangles are tracked and drawn at once. Each search pass finds the best-scoring valid triangle per window end, not just the single overall best, so two non-overlapping triangles elsewhere on the chart both surface. Aufgelöste Patterns behalten (on by default) controls whether a triangle’s drawings are deleted once it breaks out, or kept as bounded, finished segments (finalized at the breakout bar, not left extending indefinitely). Kept triangles still count for overlap purposes — a later candidate at the same spot only replaces one if it scores higher; otherwise it’s skipped.

Debug

Enabling Score-Breakdown Debug-Tabelle shows a table with the current best-in-depth candidate’s per-criterion points alongside each hard gate’s pass/fail (OK/GATE) state, plus overall pattern validity and trackability.

Enabling Pine-Logs aktivieren (Fehlersuche) (off by default) emits TCS SWING/TCS SEARCH/TCS PATTERN FIRED/TCS PATTERN REPLACED/TCS BREAKOUT lines to the Pine Logs pane, plus TCS PATTERN SKIPPED (stillborn) when a valid candidate is discarded because price already broke it by detection time, and TCS PATTERN SKIPPED (kept overlap scores higher) when a kept triangle at the same spot outscores it — useful when a symbol/timeframe combination seems to produce no signals at all, or to check why a candidate wasn’t tracked.