The review home for clouds, moving averages and the next indicator families. Open only what we’re discussing. Saved links and dated captures stay here.
Checkpoint · September 24, 2026 · New detailed workbench and captured multi-timeframe geometry. Existing Station clouds stay untouched.
New · deployed review build
Detailed oscillator workbench
One price chart and three levels of detail, sharing one viewport. Source-timeframe fans stay fixed when price bars change. This is the next review surface—not an already approved Station integration.
RSI14, Williams +100 and stochastic 14/3/3 across 3H–2W; graduated line weights, family colors, endpoint names, context zones and an optional RSI SMA14 signal band. The 50 reference is not a moving average.
62 automated checks, including mocked canvas execution; 20 live provider reads across SPY/AAPL. Browser visual acceptance and numerical TradingView parity remain open. No Station production change.
Real MCP capture · fixed snapshot
Lines that survive a timeframe change
Twelve four-hour and twelve daily BTC rails appear together. Switch the displayed candles to 4H, daily or weekly without rerunning the detections. Original segments and dotted continuations remain separate.
Includes signed distance to each nearby upper/lower rail. Unresolved historical anchors are not guessed. This is frozen Trendoscope evidence, not a live registry, parallel-channel transport or proof of predictive strength.
Source capture: 24 September 2026. Parent pattern identity and confirmation timestamps are unavailable. Do not use this snapshot for historical performance statistics.
Handoff · readable brief
Statistical evaluation & database transport
What to store, how to preserve source geometry, how to screen boundary proximity, and how to test cross-timeframe agreement without double-counting correlated patterns.
ACP screening fork compiled: zero errors; one existing-base consistency warning. Not installed or run in Pine Screener. Its ten outputs cover boundary prices, distances, proximity, continuation flags and duration.
Updated · dynamic type + dashed 8D
Cloud Workshop
Labels belong beside the latest bars, not on a price axis. Pan into history and they disappear. Reset brings them back at the exact line height, with price and percentage distance.
Nearby labels stagger horizontally, never into invented vertical price levels. Type and stroke sizes now respond to zoom; price stays thicker than every indicator. Historical pan hides latest-bar labels rather than inventing old endpoint values.
Station-derived equal-bar spacing, dominant green/red day-change price trace, line/candles/OHLC switch and horizontal pan. Wheel/pinch zoom; double-click reset.
One confirmed-daily MA ribbon across price timeframes. Each daily value is available at the next supplied session; historical joins are display geometry, not added computed samples.
Price-only autoscale; thin dashed EMA8 and optional dashed SMA100, neither adds a cloud. EMA8 is the quieter step below EMA13: width 0.6, indigo opacity 20%, pink opacity 8%. Approved hue families are unchanged.
54 local checks passed. Daily values are internally calculated over provider candles, not database MA records. This confirmed-session preview is not a developing-daily TradingView series; full provider/database equivalence is not claimed.
How this goes onto Station
Clouds are the baseline inside the existing Station chart, reusing its header and synchronization—not another chart application. SMA100 remains an optional level; the workshop Clouds checkbox is an inspection aid. No manual script copying. This preview does not change Station production; the data/session contract must remain explicit when the reviewed overlay is integrated.
Red filter at night · preserve the approved look
Keep the indigo/pink palette Alan likes under his filter. Thickness, dashed levels and text identity provide cues beyond hue. This pass makes only a bounded balance trial; no replacement night palette or OS-filter changes. Actual perceived balance still needs Alan’s filtered-display review.
TradingView · rollout candidate
Clouds Unified
EMA13 / EMA21 / SMA50 / SMA200. Three clouds, integrated labels and signed percentage distances. Widths 1 / 2 / 3 / 4. Optional dashed 100-day SMA stays separate.
52 local tests passed. Independent same-input EMA13/21 and SMA50/100/200 arithmetic passed on 1,200 daily bars each for TSLA and GOOGL. Full TradingView/provider/database equivalence is still open. Compact label defaults are in the saved source; existing chart overrides need a clean-instance check. No more TradingView feature expansion is planned.
Retained companion
Full moving-average fan
The many-line companion remains available independently of the unified cloud ribbon.
The V2 source is local and paste-ready; it has not been applied to the saved TradingView chart. It adds line labels and three subtle context zones at 68–72, 48.5–51.5 and 28–32.
32 / 32 focused tests passed. MCP pine_check: 0 errors / 0 warnings. Those results cover source structure and compilation, not visual acceptance on the saved layout.
Saved V2 preserved · compiled V3 review pending
Diagonal continuations
Upper-green/lower-red rails with thin dotted tails and no interior fill. The saved V2 chart remains intact. The distinct V3 visual candidate is source-tested and MCP-compiled, but it is not applied to the chart.
Original rails are 60% opaque and 2–4px depending on duration. Dotted tails stay 1px: 35% opacity at 20 source bars or less, rising to 55% at 120 bars. Actual saved Inputs checked. Colors describe boundary role; label direction is geometry, not a breakout forecast.
Each tail starts at the original endpoint with the same slope. That endpoint is not necessarily a confirmed break. Pattern retirement removes its tails too; 20 patterns retained. Native detection is unchanged, and still follows the chart timeframe. Locked source-timeframe transport is not yet implemented.
Pending V3 trial: original-rail opacity 45–75% and tail opacity 35–70% over the same 20–120 source-bar duration range. Larger names centered on the original rail. Pine labels have no rotation control. No continuation suppression was added; retirement still removes each original-and-tail pair.
History ceiling: diagonal calc_bars_count and retained detection state are separate limits. Raising the calculation window does not by itself retain more detected patterns.
This is a private visual derivative of Trendoscope ACP under its retained CC BY-NC-SA 4.0 notice, not a new detector or blanket permission for commercial Station use. The exact 31,534-character V3 candidate compiles through the MCP with zero errors and one consistency warning; it remains unapplied. No new minimum duration, touch threshold, nested-pattern suppression or detection filter was added.
Parallel-only · exact saved split
Parallel channels
The saved parallel review enables four lanes from two additive shared-slope families: Auto Parallel Channel confirmed/history plus developing, and SCINTILLA APC lower-primary plus upper-primary offsets. Midlines are presentation, not a third detector.
Auto Parallel Channel · sohaibnori: an open-source two-high/one-low, seven-offset chart-timeframe ladder. Active/developing sets extend right when enabled. Replaced sets end at the replacement bar plus the history offset (default −65); three historical sets are retained. That endpoint is not automatically a breakout. Source publication ↗
SCINTILLA APC offset modes: transparent upper-/lower-primary shared-slope geometry reconstructed from IQ-TRADER's protected publication description ↗. It is rebuilt through the current bar without an infinite ray. It is not a clone of the protected pivot search, scoring, touch accounting or lifecycle.
Independent-fit comparison: the separate saved chart enables TradeSymbiotic APCh HTF ↗, SCINTILLA APC independently fitted dual rails, and sunnywilson93 Auto Channel Detector v7 ↗; the four shared-slope lanes above are off. Auto Channel Detector also classifies ranges, wedges, triangles and broadenings, so the comparison is intentionally broader than strict parallel channels.
History/object audit: channel B archive 10 / 40; channel D archive 20 / 80; combined object budget 474 / 500. The audited candidate compiles at 0 errors / 0 warnings.
The maintained comparison artifact is SCINTILLA_Parallel_Channels_Additive_Stack_V1.pine. The two chart views change per-instance toggles only: no method was deleted, fused with Trendoscope or allowed to import pivots into it. No channel calculation or lifecycle changed in this registry pass.
Source audit · not equivalence certified
Pivots before patterns
Trendoscope starts from rolling highs/lows, not the symmetric confirmed pivots used by the channel scripts. Our Pivot Atom already includes rolling-extreme candidates, but its custom zigzag confirmation rules are different.
The distinction that matters
Candidate coverage is not selected-sequence parity. The local pivot union has symmetric windows, trailing extremes, flips, run completion, smoothed extremes and custom zigzag lanes. Exact anchors, equal-price handling, confirmation times and recursive selection need comparison before calling it a Trendoscope match.
Keep the raw detections. Discuss session-aware duration, distinct touches and nested-pattern emphasis as display rules before removing candidates or claiming probabilities.
Research candidates, not installed or performance validated. Audit confirmation versus anchor time, overlapping setups and out-of-sample outcomes. Check source licenses before commercial incorporation.
Deferred
Volume, pivots & Fibonacci
Separate discussions. Preserve the cloud visual grammar, but do not assume these methods have already been integrated or approved.
No new calculations or screening claims are made by this catalogue.
MCP extraction · September 24
The exact saved Trendoscope target 99awrQQB was read on BTC 4H and daily, then restored to 4H. The MCP returned 136 / 80 line primitives. Resolved endpoints cover 24 / 24 primitives, including 12 / 12 original solid rails; unresolved anchors stay excluded from the display and preserved in the evidence.
Those 24 original rails now drive the cross-timeframe workshop demonstration. This validates usable extraction for this capture—not every historical line, pattern identity, confirmation time, provider parity or database computation.
The Pine screening fork compiled with zero errors and one warning also present in its base. It is not applied. Launch/login and browser visual checks were not retested in this pass.
Latest connection-check screenshot · six-chart / 4h · September 19
Older Desktop reminder layout, not the approved unified-cloud review. Additional per-indicator/timeframe captures will be added when their exact current state is verified.
