A QUARTERLY THEORY MODEL
INSPIRED BY TRADESDONTLIE · DEVELOPED BY ANTHONY D. LAYNE
TIMING,
LOCATION,
ACCEPTANCE.
An AI trading algorithm for the compression and expansion of a range. An AI agent reads your charts as numbers, runs every VCS rule across the markets it watches, finds the setups, arms the alerts and proves each entry before calling it READY — then measures its own edge honestly. You make the final call. Fade the boundary when price rejects it. Ride the break when price accepts through it. The whole method reduces to one discipline — buy low and sell high.
VCS — 0.5R
Session Performance
Volume. Confluence. Statistics. | Last 12 months, win rate at a +0.5R target
| Session | GC | NQ |
|---|---|---|
| Asia | 67.2% | 65.5% |
| London | 69.1% | 74.5% |
| NY AM | 70.6% | 65.2% |
| NY PM | 70.3% | 63.6% |
| Overall | 69.1% | 67.5% |
Highlighted = best session per market · FINAL ENTRY · London 2:00–5:00 ET · NY AM 8:00–10:30 ET.
VCS Principle: Precision first. Consistency across sessions. Let the statistics validate the edge.
TWELVE
CHAPTERS.
VCS IN
FIVE MINUTES.
Read this page first. It covers what VCS does, what each alert means, when VCS is active, the grade on your alerts and the five rules that matter most. Every chapter after it goes deeper, and the Glossary explains every term in plain words.
IT FINDS THE LEVEL, WATCHES IT, AND PROVES THE REACTION.
VCS is an AI trading algorithm built on one idea: buy low and sell high inside a range — or ride the break when the range gives way. It reads your TradingView charts as numbers, mainly Nasdaq futures (NQ1!) and gold futures (GC1!), with GBPUSD, USDJPY and Bitcoin also supported. It finds the price levels that matter, sets TradingView alerts on them, and then checks with completed candles whether price really reacted there.
When the proof is in, it tells you. VCS never places a trade — you decide, and you click.
WATCH → EARLY → CONFIRM → FINAL ENTRY.
Each stage carries its own plan: an entry, the setup’s structural stop and a +0.50R first target. If price is already past the stop or the target, the plan reads NO_PLAN — the stop is never moved to make it fit.
FOUR WINDOWS A DAY.
| Window | Time (New York) | Days |
|---|---|---|
| Asia | 7:30 – 10:30 PM | Sunday – Thursday evenings |
| London | 2:00 – 5:00 AM | Weekdays |
| NY AM | 8:00 – 10:30 AM | Weekdays |
| NY PM | 1:30 – 3:00 PM | Weekdays |
A CONFIRM or FINAL ENTRY that fires on a weekday between 8:25 and 8:45 AM carries ⚠ 8:30 NEWS WINDOW: big economic releases such as CPI or the jobs report can move price hard in both directions. It is a warning label only — it never blocks, delays or hides an alert.
A · A1 · B · C1 · C — HOW MUCH THE BACKGROUND AGREES.
The grade is built from last month’s midpoint (PM50), the Midnight levels and the New York opening range. It never blocks, delays or changes an alert, its weights are unvalidated starting values, and it is not a win rate. Treat it as background, not as a signal.
Essentials, the default, shows each chapter’s plain story; the exact rules, checks and release notes sit folded under Under the hood. Switch to Full detail in the top bar to open them all — your choice is remembered. Every chapter opens with an In plain words box. Press / to search the whole guide.
A good first path: this page → Chapter 02 (ranges) → Chapter 04 (alerts) → Chapter 06 (breakouts) → the FAQ and the Glossary.
THE STACK.
Quarterly Theory supplies time context. Native M1 IFVG supplies the required reversal confirmation for normal routes; on GC, NQ, GBPUSD and USDJPY it must pair with a qualified reaction at the armed location (REACTPROOF/1). Structure, value and existing location sources keep their own roles.
VCS uses a small set of tools, and each one has one job.
Each row below says exactly what VCS reads from that study, and which ones VCS recomputes from its own bars.
The time layer. Splits every cycle into four quarters — accumulation, manipulation, expansion, continuation — across sessions, days, weeks and beyond. Also the single owner of session-caution: the canonical 09:00–10:30 ET window states live here (Ch 05). tradingview.com/script/nGqjV24C-Quarterly-Theory-Cycles-Oracle
Confirmed M1 inverse fair-value-gap events, read from verified native numeric outputs. VCS checks exact feed, settings, direction, event time, age, continuous coverage and invalidation. Source evidence alone never creates CONFIRM, READY or an order. Source roles and availability. github.com/cantolab/IFVG-Ultimate-Dodgy
The primary reaction proof for GC, NQ, GBPUSD and USDJPY reversals (REACTPROOF/1). VCS computes the sweep itself from completed M5 bars, with parity of 44/44 against the study. The study on the chart is therefore an optional visual that production never reads. When no sweep qualifies, the controlled fallback is a native reclaim of the armed level: a touch, then a completed M5 close back on the trade side. tradingview.com/script/9iPARtWO-Liquidity-Sweep-ATR-Envelope
Verified GCGrid levels own gold setup locations (NRA/1). VCS reads the study's own level labels — FUT-origin levels, CFD-origin levels mapped onto the futures chart by the study's own confirmed conversion (provider, spread and matched bar checked), and FastZones — and matches each against the drawn line and band. A verified level can own a BUYLOW, SELLHIGH, BREAKUP or BREAKDN setup at its exact line; the band around it never replaces the line. The Nomada reference, the WATCH approach, the tested line and the VCS entry stay distinct. CFD owns the location only — confirmation stays on COMEX:GC1!. IFVG, Gate 6, READY and structural stops are unchanged. See Chapter 08. tradingview.com/script/ZIQRZaiD-GCGrid-Psych-Levels-NomadaScalper
Volatility environment, descriptive only — never a price source or directional signal. Recent volatility of (close−open)/open returns divided by the instrument's own five-year norm, for the DAILY, SESSION and HOUR scopes, in the study's own words (HIGH · RISING, QUIET · STEADY). VCS acquires it on H1 only — a supervised context job reads a dedicated read-only chart at startup and every half hour, and the top-down sweep reads it on H1 — and shows that same hourly snapshot on M30, H1 and H4, labelled source H1. It respects the n ≥ 201 sample gate; a reading is current only inside its own dated window and within 35 minutes of being read. HIGH is not bullish and QUIET is not bearish. See Chapter 08. tradingview.com/script/K6xhBPjw-NS-MARKET-REGIME-NomadaScalper
Six zones own gold setup locations (NRA/1, September 29, 2026). D− PIVOT, D− MID, D− EXT, D+ PIVOT, D+ MID and D+ EXT are read from the study's own labels and boxes — never rebuilt. A fade tests the near boundary, a breakout the far boundary, and the full interval stays on record. The zone signs are names, not sides: price decides above or below. A touch is not a rejection, and an EXTREME never forces a reversal. NO REVERSAL and OPEN GAP remain context. See Chapter 08. tradingview.com/script/k6LBPeO2-GC-Xau-Daily-Reference-Map-NomadaScalper
London High and London Low become NQ locations (ALNCTX/1, September 30, 2026). VCS recomputes the Asia and London ranges from its own NQ1! bars, so the study is not required on the chart. After the 08:00 ET lock, the London levels not yet swept are the next NQ location tier, after the Reactive NQ Grid and before the native levels. Every percentage shown is a historical frequency labelled hist, never a prediction. See Chapter 08. tradingview.com/script/KNTFWlm2-ALN-Sessions-NQ-Stats-x-CantoLab
The grid owns Nasdaq setup locations first (RNQG/1). VCS reads only what the study draws on your NQ1! chart: the R GRID U1, U2, L1 and L2 zones, the Upper and Lower MID P/E lines and the MID band, each checked against its drawn line or box. The study's method stays private and is never rebuilt. A zone or line can own a BUYLOW, SELLHIGH, BREAKUP or BREAKDN setup; price decides the side, never the U or L name, and MID is only a target reference. A fresh BUY NS, SELL NS or NS React marker at the armed level also counts as the reaction proof, and the IFVG is still required. Then come the London levels, the Midnight levels and the native levels. tradingview.com/script/9ZPVadau-Reactive-NQ-Grid-NomadaScalper
The midnight hour becomes a location and a target (MRS/1). VCS recomputes the study from its own bars, so the study is not required on the chart: the full 00:00–01:00 ET hour gives a Midnight High, a Midnight Low, their exact midpoint (MID) and the Midnight Open. From 08:00 to 16:00 ET the High and Low own setup locations after the Reactive NQ Grid and the London levels on Nasdaq, after the Nomada levels on gold, and first on GBPUSD, USDJPY and BTCUSD. While the day has not touched them, MID and the Midnight Open can be the structural target. Each percentage is how often, across the completed sessions of the loaded history, the 08:00–16:00 window came back to that level — printed as hist with its sample size and 95% interval. See Chapter 08. tradingview.com/script/jK0YiUpG-Midnight-Revisit-Study-NomadaScalper
The New York opening range, and what usually happened after its first breakout (TBR/1). VCS reads only what the study draws on its background chart — never your chart — inside each window instance on weekdays: by default the NY AM instance, the 09:00–09:30 ET range read from 09:30 to 10:30 (and the NY PM instance, 13:00–13:30 read to 15:00, once you seed it), the direction it first broke, and the study's continuation and reversal percentages for that breakout from its own history. When one side reaches 60%, it becomes one vote in the context grade on CONFIRM and FINAL ENTRY; otherwise it abstains and says why. Its zones are reference only and never move a stop or a target. The study's method stays private, and your seed copy never leaves your machine. See Chapter 04. tradingview.com/script/4r56IeAU-Time-Based-Ranges-The-Daily-Profiler
IFVG CONFIRMS.
PROOF STILL DECIDES.
A marked historical candle is not the time the system knew the event. Only completed native observations with verified source identity can be used.
Under the hood2 technical notesIFVG · Normal Reversal Confirmation · Installation & Operating Limits
QUALIFY THE EVENT, THEN REQUIRE LATER PROOF.
The private source uses native M1 standard candles. The current preset is Single, Neutral, 15-bar grace, swing-price filter ON, sweep and higher-timeframe overlap filters OFF. Visual history, hiding losing setups and simulated trade outcomes never control machine evidence.
VCS admits accepted events at their confirmed known-at time, tied locally to the exact setup epoch and branch. Maximum event age is 20 minutes; native close and observation freshness are 90 seconds. A subsequent confirmed close strictly through the gap's opposite edge permanently invalidates that event. Equality alone does not.
Location can arm while confirmation is pending. Required evidence must satisfy the route's original WATCH/activation boundary. Full retest requires distinct return and hold bars; reaction hold requires proof after all confirmation prerequisites. The exact Range Mode ARM/eligible full-retest exception remains source-optional. Qualified CONFIRM and reaction hold require IFVG.
Use the private VCS IFVG evidence companion with matching saved source, input fingerprints and native readback. A copied script is not a verified installation on another account. Preserve Quarterly Theory, unrelated studies and managed price alerts.
The source reader uses the existing chart owner and restores the prior interval and visible range. Ambiguous panes, wrong feed/settings, stale data, insufficient coverage or source overflow return unavailable. Keep the local host awake and connected. The local reader needs no webhook; the separate REACT/2 cloud transport uses a verified HTTPS receiver.
Breakout confirmation remains structural. The indicator cannot change branch geometry, score or risk. The September 15 NQ installation evidence is retained. September 20 native BTCUSD readback and bounded independent IFVG comparisons are recorded separately in the REACT/1 release report. Statistical status remains UNVALIDATED.
VCS RANGE
Mean reversion at a defended boundary. Five steps, in order — location, structure, the coil, defense, targeting. Skip one and it is not a VCS trade.
A range trade fades the edge of a range: buy near the floor, sell near the ceiling — but only when every step lines up.
The entry is timed on the 5- and 1-minute charts, and the stop sits where the setup is proven wrong. Skip a step and it is not a VCS trade.
FIND THE RANGE. SET THE BOUNDARIES.
Every VCS RANGE setup begins with compression detection — the driver's compPctile / coil3 read top-down across H4 / H1 / M30 / M15 to locate where compression is active. On the higher timeframe, the compression zone marks where price is coiled and ranging — that zone is your range location and your primary setup.
Once the compression zone is confirmed, its top and bottom become your boundaries: range high (resistance) and range low (support). Confirm at least one boundary aligns with a defensible structural level — the prior day's high, low or midpoint, the current day's extremes, or a session extreme — all computed from raw OHLC, no indicator required. These ARE the armed edges, with one guard: a level closer than 0.25× ATR is coil noise, never a boundary — a stop there has no room to be right.
WAIT FOR THE SECOND COMPRESSION.
Once price reaches the range high or low, drop to M30 / H1 / H4 / H8 / D and wait for a new compression zone to form at the boundary. This is price coiling a second time — now at the exact level you are trading. Compression describes the available structural evidence; it does not supply a calibrated probability of rejection back toward the range center. No coil = no setup: price still free-falling into a level with compression expanding is a flush in progress, not a trade.
Then read the boundary with its location context (Ch 03): the psychological grid and the existing liquidity read. Strong rejection at a boundary that is coiled and structurally defended is the highest-quality fade read this system produces. The same location accepted through instead is breakout evidence — never an automatic fade.
IS THE LEVEL BEING DEFENDED?
Defense is read off completed bars at the boundary: rejection closes that leave wicks beyond the level and bodies back inside it, failed acceptance (a poke through that cannot hold a close), and the retest that holds. A level being defended prints refusal you can see on the bar itself; a level being absorbed prints acceptance — bodies closing beyond and staying there.
The market-state engine (Ch 03) supplies the label over the top: BULL, BEAR or NEUTRAL, computed from prior-day range position, swing structure and completed-close acceptance. A fade with the state is vetted; a fade against it is honestly tagged counter-trend and judged on the live tape only. The label never blocks a side — NEUTRAL is a first-class answer that arms the coil both ways.
FADE THE EDGE. TARGET THE CENTER.
Your take-profit is always a price level, never a timeframe. Mark it before you enter.
ENTER PRECISE. TARGET THE CENTER.
Execute on M5 / M1 for a precise entry. Target the center of the range. Place the stop just beyond the range boundary, the tightest level that still sits past genuine invalidation — structure decides where it belongs, never a ratio. Your TP is always a specific price, never a timeframe.
VCS RANGE does not target an arbitrary ratio. The system identifies the defended boundary, the logical invalidation and the legitimate reversion objective; once Gate 6 establishes the actual entry, the resulting R:R is calculated: risk = entry to structural invalidation, reward = entry to the structural objective, R:R = reward ÷ risk. Sometimes that is under 1R; sometimes it is 2R, 3R or more. Report the geometry honestly and never alter structure to manufacture a ratio — never tighten a legitimate stop or invent a farther target to make a trade display prettier numbers. Your position size, not your stop, is where you control risk. The center is targeted because it is the structural reversion objective — not because it happens to equal a predetermined R multiple.
SECURE THE FIRST MOVE. LET STRONG TRADES RUN.
The base profit objective on every VCS trade is +0.50R — 0.5:1 reward:risk. It is the first meaningful favorable move, and securing it is the default: taking the full profit there is valid, banking a partial there is valid, and holding beyond it is equally valid when the trade keeps earning it. This changes when you monetize the move — never how the trade is built. The structural objective still defines the trade's geometry, and R:R stays an output of that structure.
Under the hood1 technical noteSession Context
Earlier replay-era measurements suggested efficiency clustered into the London/NY overlap and late-NY windows, mid-week. Treat that as a historical research prior, not a rule — the current measurement layer (Ch 09) has not re-validated any session or weekday claim at its evidence bar, and old measurements never become permanent doctrine. Session state still matters as context (Ch 05); it sizes attention, never entitlement.
MARKET STATE &
AUCTION CONTEXT.
Two layers read the market around a setup: a deterministic market-state engine computed from raw price, and a context layer — the psychological grid — each with an authority label it cannot exceed. Bias is now a label on the trade, never a gate in front of it.
Before a setup counts, VCS reads the market around it: is price trending, balanced or undecided, and which source owns this level? These reads label a setup — they never create or cancel a trade on their own.
Most of this chapter is reference for experienced users, so it sits folded under the hood.
THREE STRUCTURAL VOTES. THREE HONEST ANSWERS.
The engine reads three things any trader could verify on a bare chart: where price sits in the prior day's range · swing structure (higher highs/lows or lower) · completed-close acceptance (which levels have bodies closed beyond, not wicked). The votes sum to one of three states — BULL, BEAR, or NEUTRAL — and NEUTRAL is a first-class answer, not a failure: it means the tape has genuinely not picked a side, and the coil arms both ways.
FEWER LAYERS. CLEARER ROLES.
Location comes from structure and the psychological grid, each at its own authority ceiling. Both branches at a boundary stay armed and observable, and completed closes decide which one proves itself: a level accepted through strengthens the breakout, never the fade.
Under the hood12 technical notesOTE — Retracement Geometry · GATE 6 — The Execution Proof, Branch-Aware · Evidence Integrity — Count Witnesses, Not Echoes · One Trade, Many Labels · Sessions Are Market-Specific · One Command, Two To Five Minutes · Full System Audit — Source And Evidence Integrity · 28th Principal Review — Evidence Integrity · 25th Principal Review — Verified Boundaries · Session Truth — One Clock, One Classifier · Alert Names — Where, And What Matters Now · PSY — The Psychological Grid
A measurement layer, not a signal. From the current confirmed impulse — a swing low to a swing high, or the reverse, anchored by deterministic completed-bar pivots that never repaint — the engine marks the 62–79% retracement zone with a 70.5% reference that is tracked, never privileged. The research question is narrow: when a VCS reaction happens inside that zone, does it pay better than the same reaction outside it? The answer waits for governed samples. Until then: OTE is not an entry, not required, never replaces the reaction, never replaces the delivery confirmation, never replaces the execution proof — and a zone touch with no VCS reaction is nothing at all. No impulse simply reads UNAVAILABLE, which penalizes nothing. Overlapping geometry stays one judgment — a structural level inside the zone beside a psychological grid line is a stronger location, never three confirmations.
Gate 6 has always demanded one thing: proof before entry. That demand is now stated per branch. On the breakout side — BREAKUP and BREAKDN — the classic retest-hold remains the first proof, with a distinct return bar followed by a holding close — and since August 30, 2026 it is joined by two continuation classes the same funnel measurement demanded: the shallow-retest hold — price displaces, pulls back toward the broken level without ever fully returning, and a completed bar closes back onside — and the no-retest continuation — displacement established and then sustained by a further completed close, with no meaningful return at all. A full retest, once it prints, still owns the episode; a reclaim close still kills the branch; and no delivery confirmation is borrowed — breakouts stay structurally driven. On the reversal side — BUYLOW and SELLHIGH — measurement showed the strict retest was refusing most confirmed fades that simply never returned, so a second proof class now stands in production: the reaction hold — a fresh completed bar closing on the trade's side after the fire, with zero closes back through the fired level. Two classes, one standard: the market must demonstrate, on completed bars, that the level is holding. The trigger bar can never be its own proof, an invalidation close kills every class equally, and the no-chase line is absolute — price beyond the reach ceiling is a missed trade, recorded honestly, never chased. Promoted to production on August 29, 2026 by operator charter — and paired with the same repair upstream: the canonical BUYLOW/SELLHIGH confirm is itself dual-proof now. The armed alert cross remains one path; the other is the completed reversal evidence the narrow cross can miss — penetration, a completed rejection close on the family's own confirm timeframe, and a accepted, causally current native M1 IFVG — recorded through the same fire recorder, with the first valid proof winning and the second kept as evidence only. Honesty stands: none of the new cohorts — reaction hold, shallow-retest hold, no-retest continuation — is a validated edge; the Chapter 09 measurement bar still owns that verdict — and every CONFIRM and READY now names its proof (confirm_proof · ready_proof) so the ledger can render exactly which path earned it.
VCS counts independent evidence, not indicator quantity. When a psychological level, a retracement zone and a structural line all sit in one reaction zone, that is one piece of geometric evidence with rich provenance — three drawings of the same place, not three witnesses. A location carrying five overlapping drawings of one level is not five confirmations, and treating it as five is how confidence gets manufactured instead of measured: ask what independently agrees, never how many things are drawn there.
One physical market move can wear many badges: a location reaction, a pre-alert, an execution-proof candidate, a final branch, a real fill. Those are stages of one story — never separate trades. The books therefore count one root opportunity, one branch attempt inside it, and one execution when money actually moved; partial fills and scale-outs are details of that one position, and hitting +0.50R then +0.75R then +1.00R is one trade passing milestones, never three wins. Real results, theoretical shadows, and replayed reconstructions live in separate populations that never pool by default — every published number names which population it came from, and a headline win rate with no population label is not a VCS number.
Gold's morning is not the Nasdaq's morning. Session performance is earned per market, per session — a strong GC NY-AM says nothing about NQ NY-AM, and a pooled "futures session win rate" is forbidden. The original three-session study also carried a sampling honesty problem: its historical arm points froze at 03:00 and 09:00 ET, so London's first ninety minutes and the whole fresh afternoon were structurally unmeasured. The corrected full-window study arms a fresh cycle at each official session start — 01:30, 09:00, 13:30 — with everything else frozen, and the old study stays untouched as the record of what it actually measured. Early findings (all below the audit sample floors): the recovered early-London window is net-negative on both markets at the first-burst target, Gold's edge stays concentrated in NY-AM, and the Nasdaq's only positive fresh window is the afternoon. Session context tells you WHEN the system has historically paid — it never picks a direction, never replaces confirmation, and never blocks a valid setup.
The normal workflow is one command: turn the machine on, open the assistant, type EXECUTE GC, EXECUTE NQ, or a configured analysis-only pair such as EXECUTE GBPUSD. A deterministic conductor now routes the command, chooses FULL or FAST_REFRESH, performs exactly one immutable acquisition, grades once, arms through the existing owner, reconciles, and records one end-to-end timing row. This removes the 21–41 shell-call detours that the audit measured at 4.7–5.6 minutes of coordination per run. The chart gate checks TradingView's internal ticker and timeframe instead of a human legend or timeframe-dependent DOM bar nodes, and each feed is selected once. The first non-arming GC FULL canary completed all 10 requested chart states with zero gaps in 40.642 seconds, down from the pre-fix observations near 135 seconds; one canary is evidence the repair is active, not a latency distribution. A small source-hash-pinned live kernel replaces hundreds of kilobytes of ordinary startup context. A healthy analysis targets two to five minutes — as a service target, never a permission slip: no timeframe, confirmation, gate, risk check, or authority changes. The system does the same trading work with less waiting around it.
Exact source before READY. Retest checks require the intended analysis ticker; an NQ setup cannot advance on an MNQ or similarly named pane. Connection recovery verifies the TradingView hostname and closes failed sessions. Existing confirmation, structural stops, no-chase and risk checks remain in force.
Outcome calculations exclude bars beyond the observation time. Model qualification now includes shared outcome and statistical code, so changes require fresh qualification before model influence. The audit provides software checks; live alert delivery and statistical edge require their own evidence. Branch notifications retain their original details. See docs/PRINCIPAL_REVIEW_2026-09-12.md for findings, coverage and release receipts.
Principal review 28. XGB remains in COLLECT with normal baseline alerts and no qualified market model. Training now requires a verified, complete evidence record; late outcomes cannot be treated as previously known, and corrections return model influence to baseline until revalidated. Fresh-package checks now include the full test runner’s required files. Existing dashed alert markers, confirmation, READY, stops, targets and hard risk limits retain their current contracts. Review findings, scores, tests and remaining live-verification limits: docs/PRINCIPAL_REVIEW_2026-09-08.md.
Exact alert ownership, confirmation, Gate 6 and the no-chase ceiling remain in force. UT stays OFF; its private Pine verification is separate from production activation. The review records full offline regression and package checks, with live-webhook and statistical-validation limits kept explicit.
Four official intraday campaigns, stated once and owned by the runtime: ASIAN 19:30–22:30 · LONDON 02:00–05:00 · NY AM 08:00–10:30 · NY PM 13:30–15:00 — New York wall clock, daylight-saving safe, boundaries half-open: 14:59:59 belongs to the afternoon campaign, 15:00:00 does not. The current campaign is a machine fact: one classifier converts the timestamp, every consumer adopts its answer, and nobody — operator or assistant — decides the session by glancing at a clock. Outside the windows is a real state, labeled and never blocked; an unreadable timestamp is an error, never quietly "outside". One honesty rule worth framing: the machine's ledgers timestamp in UTC while the campaign windows live in New York wall time — a stamp ending in Z runs four or five hours ahead of the wall, which is exactly how a correct label can look wrong at a glance. The operator surface now prints the Eastern wall time beside every campaign so the two clocks can never be confused, and the host machine's own timezone — whatever it happens to be set to — has no authority at all.
An alert title answers two questions and no more: where is price reacting — the market and the structural source (GC1 PDL, NQ1 CDH) — and what VCS state matters there: WATCH when price arrives at the battlefield, or the branch itself — BUYLOW · SELLHIGH · BREAKUP · BREAKDN — once a reaction earns the name. The machine's internal chronology (touch, print, penetration, rejection, acceptance) still runs every lifecycle exactly as before — it is simply no longer a title. WATCH is a notification, never a fifth family; a generic "break" or "reject" is an event class, never a trade. And the separation is structural, not stylistic: machine identity lives in the alert's tag, never in its visible words — renaming a title can never change what the state machine believes, and alerts armed under the old names keep them until they are naturally replaced. Fewer words on the phone, the same intelligence underneath.
WHERE IS PSYCHOLOGICALLY OBVIOUS PRICE — ON GOLD AND NASDAQ?
Round numbers get respected with a regularity worth measuring: every 100-unit multiple, every 250-unit multiple, and the 50% FastZone midpoints between them. PSY/1 runs the GCGrid — Psych Levels arithmetic (NomadaScalper supplied v3.1.0 native geometry) natively — source-faithful on Gold, and since August 20, 2026 ported to Nasdaq futures on an explicit user-approved charter — the indicator itself is optional wall decoration, never a runtime dependency. The operator term for the mapped area is the PSY S/R Decision Zone: a psychological support/resistance area where VCS waits for confluence and confirmation before any trade is authorized. The zone says pay attention here — it never says price is guaranteed, proven, or required to react. The shorthand: PSY = where · IFVG = delivery confirmation · Gate 6 = execution proof.
Every layer in this chapter carries a ceiling it cannot exceed: MSE labels, and PSY marks psychological location on its two configured families — Gold and Nasdaq futures. Promotion — a context layer earning a score, a gate, an armable level — runs through the Chapter 09 bar: adequate samples, the full audit, explicit sign-off. Never a config tweak, and never silently. The regression suite pins the boundaries: context remains without independent score or execution authority.
BRANCH.
FORK.
LADDER.
The lifecycle ties every alert together. WATCH brings attention to a qualified decision area. EARLY is a heads-up — get ready, not an entry yet. CONFIRM is the primary entry alert. READY alone marks execution authority—and VCS never turns it into an order.
Under the hood1 technical noteAlert Activation · September 10
ALERT ACTIVATION · SEPTEMBER 10
The GC/NQ installation failures occurred while five newly created alerts were still inactive. Their prices and candle-close conditions matched. A comparison between the provider and local creation times incorrectly skipped the short activation wait, so those alerts were withdrawn.
VCS now verifies that the exact alert was created by the current operation before allowing the same five-second wait. An existing, unknown or stopped alert cannot gain restart permission. Every condition, ownership, freshness and proximity check still applies. Earlier withdrawn alerts require explicit restoration and fresh validation.
Completed analysis is separate from installed alert coverage; neither alone establishes an entry-ready trade. TradingView remains the phone notification channel, and local CONFIRM/READY stays desktop-only. Enabled settings do not prove phone receipt. Triggered one-shots remain inspectable for 24 hours, with their exact event saved before cleanup.
BREAKOUT / ACCEPTANCE
Market accepted the price. Trade the break — after the retest holds.
RANGE / REJECTION
Market refused the price. Trade the fade — on the confirmation close.
EVERY BOUNDARY IS ARMED BOTH WAYS.
Each valid edge carries two branches: a REJECT fade (an M5 confirmation close back through the edge → fade to the range center) and a BREAK expansion (an M15 BODY close through the edge → stalk the retest). The bot arms the valid ones automatically at the end of every analysis run. A fire consumes only that alert; every other still-valid branch remains armed until its own lifecycle rule resolves it.
THE CANDLE CHOOSES THE PATH.
Under the hood4 technical notesCompleted Daily Levels · Live Setup Review · Bypass Safety Coverage · READY Integrity
ONE VERIFIED LEVEL THROUGH ARMING.
A forming candle can extend the daily high or low shown on the chart. New VCS daily branches use the completed-bar level that their causal evidence can verify. The new extreme becomes eligible after a fresh acquisition confirms the candle has closed.
This repairs the GC disagreement between a proposed 4381.5 level and its completed 4380.0 anchor. Initial selection and registration now share the same completed source levels. Live price, ATR, proximity limits, candidate priorities and confirmation rules remain unchanged. Existing setups keep their frozen levels.
Missing evidence and real price mismatches still block arming. An override cannot waive causal validation. This repair adds no READY authority. Verification and release details: docs/DAILY_ANCHOR_RELEASE.md.
SEE WHAT VCS IS WAITING FOR.
Run your normal EXECUTE GC command. Its VISUAL/1 link opens the exact captured setup in the local Decision Dashboard: completed native candles, frozen source and trigger, structural stop and recorded target. The condition panel explains the recorded branch state and next required proof.
Actual WATCH, EARLY, CONFIRM, retest, READY and terminal events appear as their records arrive. Pause at any observation, inspect an event, then return to live. Known-at-time replay excludes later evidence; Event chronology marks late arrivals. Export a self-contained review for later inspection. Missing evidence is shown as unavailable.
These detailed visuals are in the local dashboard. Native TradingView level markers and cloud phone alerts remain separate. The page never creates an alert, changes a level or authorizes an order; READY still is not ENTERED. Existing collection cadence and an awake connected host govern freshness. The two-second target measures captured evidence reaching the page, not tick-by-tick price updates.
Open independently with node vcs_dashboard.mjs --serve --open. Disable capture with VCS_VISUAL_REVIEW=OFF, or keep capture without opening a browser using VCS_VISUAL_OPEN=0. Contract and release evidence: docs/LIVE_VISUAL_REVIEW_CONTRACT.md and docs/LIVE_VISUAL_REVIEW_RELEASE.md.
WHEN PRICE TURNS AWAY OR ESCAPES.
For a qualified active branch, BYPASS observes an approach followed by departure, or a crossing beyond its outer safety rail. Lower near-misses warn UP; upper near-misses warn DOWN. An exact touch, an unqualified reaction and an unknown path stay distinct. Existing branch events are checked before a missing-response warning.
Safety rails freeze the decision ATR: 0.25 ATR approach, another 0.50 ATR departure, 0.25 ATR outer escape and 0.10 ATR reset hysteresis, rounded to ticks. The current observer samples through the connected host; a fast move between samples can be missed. Cloud safety alerts and phone delivery are not installed or verified.
Anthony may see a desktop VCS SAFETY notice marked NOT AN ENTRY. The existing owner reassesses the affected scope under its normal validation and repair permissions. A safety notice cannot create READY, change a stop, cancel a sibling or authorize an order. Read coverage with node vcs_bypass.mjs --status. Details and rollback: docs/BYPASS_CONTRACT.md.
THREE ALERT STAGES. ONE ENTRY-PROOF BOUNDARY.
WATCH draws attention, EARLY supplies direction, and CONFIRM validates the branch. READY records completed post-trigger proof for that exact branch; VCS places no orders. Current notifications remain enabled. This repair does not remove READY or change its supported retest, reaction and continuation proofs.
Full-retest proof keeps its actual candle-close time, so a late observation cannot look newly qualified. The existing 2.5 ATR no-chase ceiling applies to the holding close and current price. Chart evidence must match the exact analysis ticker. If a completed retest and hold are already available, the local observer records both legal transitions in the same pass. Since September 30, 2026 every READY notice states when its proof bar closed and how old it is; a proof older than max(5 min, 2× its timeframe) is marked STALE — it describes what already happened and is not a fresh entry.
Historical records stay intact. Software verification, loaded observer health and phone receipt are separate facts. See docs/READY_INTEGRITY_RELEASE.md.
WATCH FINDS THE AREA. EARLY OWNS THE DIRECTION.
Attention: WATCH · Direction: EARLY · Validation: CONFIRM · Final entry: READY, shown as FINAL ENTRY. EPRIO/2: CONFIRM is the primary entry alert; EARLY is a heads-up.
WATCH: “Price has reached or is approaching a qualified decision area.”
EARLY: “HEADS-UP · not an entry yet — VCS now favors a direction.”
CONFIRM: “The existing EARLY thesis has received stronger validation.”
READY: “The same thesis has completed VCS post-trigger proof; execution checks still follow.”
ROTATION: “The prior directional thesis failed or yielded to an opposing EARLY thesis.” The selected manual Scout is independently optional and defaults on to preserve existing preference.
| Stage | Fires when | Human delivery | Meaning |
|---|---|---|---|
| ① WATCH/1 | M1-resolution feed, intrabar at 0.50·ATR inside the structural level — not an M1 close | App + toast on every created WATCH; Scout selection is independent | Optional location attention. One eligible WATCH may include the existing capped manual Scout packet. It is not an EARLY substitute. |
| ② EARLY | First M1 close through the level | Heads-up (app + toast) | Opens one durable directional thesis. Manual-only; cannot create READY, intent, or an order. |
| ③ CONFIRM | The M5 confirmation close (fade) or M15 body close (break) | Same-thesis app + toast | Validates the existing EARLY thesis. Gate 6 and no-chase still govern READY. |
WATCHLOAD/1 health: the resident watcher publishes a fresh heartbeat containing its loaded runtime-closure fingerprint and the required VCSL CONFIRM, VCSP WATCH, VCSR EARLY, setup-epoch parser and EARLYTHREAD PRIMARY capabilities. A stale or mismatched runtime exits for supervised reload; a live PID by itself is never accepted as proof that alerts are being observed.
Under the hood3 technical notesStateful Alert Registry · One Level, One Decision · GATE 6 Routes · Sweep Hold
THE SETUP SURVIVES THE ONE-SHOT ALERT.
Causal scope diagnostics. A causal-scope refusal means a proposed branch could not be tied to complete confirmed source evidence. VCS reports the specific observation or branch-source failure, including mismatched prices, instead of treating every refusal as missing history. The September 8 precision repair handles exact half-tick midpoint differences consistently in explicit arming and resident reassessment. It does not move triggers, stops or targets; real mismatches, forming-only sources and stale or incomplete evidence remain blocked. The original GC run lacked detailed evidence; a subsequent read-only audit reproduced the numerical defect. See docs/CAUSAL_SCOPE_RELEASE.md for verification and limits.
Analysis and alert transport. Completed analysis remains visible when downstream alert installation is blocked. EARLY and CONFIRM use frozen horizontal-line conditions so TradingView preserves M1 and M5/M15 close timing; WATCH remains an intrabar price alert. Newly created instances receive a bounded activation check under the existing transaction lock. Stopped, fired, mismatched, expired or unavailable alerts are never made valid by waiting. Missing native price evidence is reported separately from an actual proximity breach. Historical missing projections still require explicit restoration and fresh validation. See docs/ALERT_TRANSPORT_RELEASE.md for live readback evidence, release checks and limits.
Native Alert Line limitation. TradingView currently does not render its native white dashed Alert Line for VCS candle-close alerts built from fixed drawing-line conditions. Enabled visibility settings and chart source linking do not overcome that limitation. The amber MARKER/1 drawing is a separate display aid. Use node src/cli/index.js alert line-status to read native support, visibility, source and timeframe diagnostics. RENDERABLE means renderer eligibility; it does not prove visual appearance or alert delivery. VCS preserves candle-close timing and all trading controls.
Visible alert levels. VCS now draws a locked amber dashed line at each verified pending alert level. Branches sharing an exact feed, setup epoch and price share one label. These markers are display only: they do not create alerts, READY or trades. They appear on matching source charts, including other timeframes, and the connected watcher refreshes them after chart changes, alert consumption, suspension or expiry. The drawings are unsaved and rebuilt from current ownership after reload. Marker visibility and alert installation are reported separately. Use node vcs_alert_markers.mjs --sync to refresh drawings without recreating cloud alerts. Global drawing visibility and the chart viewport remain your settings.
All branch notifications · BRANCHNOTIFY/1. Every created WATCH, EARLY and CONFIRM branch alert now has Notify in app and Show toast notification enabled, including ordinary MACHINE WATCH. Older mute preferences cannot suppress these cloud notifications. Optional manual Scout eligibility remains separate. The horizontal-line condition represents the fixed price level and lets EARLY/CONFIRM wait for their configured candle close; switching to a regular price alert changes the timing to intrabar. Actual iPhone receipt still requires device verification.
Persistent cloud alerts · ALERTLIFE/1. Branch app notifications now begin with the original branch details; the persistence notice has been removed from WATCH, EARLY and CONFIRM message boxes. Cloud notifications now have a separate lifetime from the local setup. New alerts use open-ended duration when TradingView verifies entitlement; otherwise they use 58 days. Session expiry and PC sleep do not themselves expire these cloud alerts. After setup validity ends, an unfired retained alert is a REFERENCE ONLY notification requiring fresh assessment; it cannot create CONFIRM, READY or an order. Invalidation, consumption, supersession, suspension and explicit stops remain effective. Local monitoring still needs an awake connected computer. Use node vcs_arm.mjs --persist-alerts --dry-run to inspect exact active-instance migration; stopped or unverified alerts are refused. True open-ended duration requires an eligible TradingView plan and remains subject to its inactivity rules. On iPhone, enable TradingView notifications and immediate delivery, allow it through Focus, and verify actual receipt. See docs/ALERT_LIFETIME_RELEASE.md.
Alert priority · EPRIO/2. CONFIRM is the primary entry alert; EARLY is a heads-up and stays the XGB learning stage. Why: STAGEENTRY/1 backtest (1-year M1, +0.50R exit, 1× costs) net R — CONFIRM −0.105 GC1 · −0.043 NQ1 · −0.238 all vs EARLY −0.200 · −0.131 · −0.365. READY remains the proof boundary. The new XGB cohort begins prospectively; original records are preserved. Actual outcomes, validation and explicit model activation are required. Normal TradingView phone alerts remain unchanged; a qualified model may add a desktop EARLY quality update. It cannot cancel a phone alert or create READY. The resident prevents automatic system sleep while running; display sleep and deliberate sleep remain available. Set VCS_ALLOW_IDLE_SLEEP=1 to opt out. The branch-tracking repair preserves a surviving token when its sibling retires. Use node vcs_xgb.mjs health; see docs/EARLY_PRIORITY_RELEASE.md.
Execution outcome reporting. Zero new branches can mean existing required projections are still verified and active. PROJECTION_RESTORATION_REQUIRED means previously installed alerts are missing, including alerts withdrawn after a failed readback. Repeating EXECUTE does not restore them. VCS reports the installation counts and exact epoch, branch and role; new installation records retain which fields failed verification. An explicit restoration request still requires fresh validation and cannot revive a terminal setup. This reporting applies across supported instruments and both FULL and FAST acquisition paths; it creates no READY authority.
Every new owned alert carries a setup epoch. ALE rows and committed alert plans are durable authority; TradingView instances are reconstructable delivery projections. If WATCH shows Stopped — Triggered, the location, time, price and candidate branches still exist in VCS state. Under AUTOREARM/1, resident and scheduled reconciliation report missing projections without recreating them by default. Run node vcs_arm_reconcile.mjs --restore-projections for an explicit one-shot restoration, or set exact VCS_ALERT_AUTO_RECREATE=PRODUCTION for persistent autonomous repair. Every authorized repair uses the shared ATX lock and a fresh under-lock inventory; foreign and legacy alerts remain untouched.
EARLYTHREAD/1 adds one deterministic thesis identity per setup epoch and branch. Duplicate EARLY observations reuse it; CONFIRM, READY, and terminal states update it; provider arrivals are ordered by canonical event time. An eligible opposing EARLY becomes one linked ROTATION update. Post-READY rewind fails closed. The first valid branch selection still suspends only its canonical opposite; unrelated alerts survive. MISSED — DO NOT CHASE remains terminal until a new setup epoch.
From this release forward, VCS stores exact native WATCH, EARLY, CONFIRM, READY, terminal, and rotation observations for deterministic shadow comparison. The five-year replay could not prove EARLY expectancy because those native events did not exist historically; VCS does not fabricate them from OHLCV. Shadow telemetry is research-only.
BOTH BRANCHES STAY ARMED. ONE CHAIN DECIDES.
A reversal and its breakout at the same price — BUYLOW and BREAKDN at a low, SELLHIGH and BREAKUP at a high, on the same timeframe and campaign — form one level case. Both stay armed and visible. When one branch's chain starts (its CONFIRM, or on a reaction-proof setup its reaction at the level), the other is suspended, not retired: it stays armed and on the chart, and nothing about it is deleted.
The first chain to finish wins. If that chain reaches READY, the level is resolved and the sibling ends. If it fails under its own rule — a completed close back through — the level is handed to the sibling once per campaign, and the sibling must complete its own full chain: its own CONFIRM, its own reaction proof and IFVG where required, its own Gate 6. A second failure closes the level. A move that ran too far (MISSED — DO NOT CHASE) never hands anything over. The level is resolved only by what the market does. Statistics may influence branch preference and directional weighting, but never override structural qualification, reaction confirmation, risk controls, or execution authority.
What you see. One VCS DECIDING line when price first meets an armed pair; the READY gains one line — Level 21450 M5: BREAKDN resolved the level · BUYLOW suspended; and one VOID line if that READY dies early. The failed-breakout reversal that used to be impossible once a breakout fired is now a normal, fully proven setup.
SWEPT, NOT KILLED. HELD AWAY, NOT RETESTED.
Gate 6 has two reversal routes. A confirmed BUYLOW or SELLHIGH reaches READY through the retest that holds (RETEST_HOLD) or through the reaction that holds away without a second retest (REACTION_HOLD): a fresh completed close on the right side, no close back through, and still inside the 2.5·ATR no-chase ceiling. Whichever completes first wins, and the level still gets one READY. On setups armed from October 3, 2026 the READY says which route it took — via REACTION_HOLD or via RETEST_HOLD.
A shallow sweep holds the reversal instead of killing it. Two structural rules used to end a reversal before it could confirm: a new day extreme beyond its level, and a wick reclaim followed by a close beyond it. When that move went no further than a quarter of an ATR beyond the level, the reversal is now held — still armed and on the chart — and a HOLD line tells you so. It comes back only through a new qualified reaction and the full chain; that READY reads via SWEEP_HOLD. A deeper move, a second sweep, or a held reversal that never revives ends exactly as before, with one VOID line.
WATCH, EARLY, CONFIRM, FINAL ENTRY — EACH WITH ITS OWN PLAN.
You enter where the timing is right, not only at READY. Every stage now carries its own plan: entry at the close of the last completed bar when that stage fired, the setup's structural stop (the same price at every stage), +0.50R from that entry, and the STRUCT target. WATCH names no direction, so it is planned both ways. If price is already past the stop or the target, the plan reads NO_PLAN — the stop is never moved to make it fit.
READY now reads FINAL ENTRY. It is the latest and most filtered entry option, not the only one: FINAL ENTRY · via retest or FINAL ENTRY · via reaction hold. Every stage is graded live as a STAGE OUTCOME (not your fill), and the dashboard shows the result by entry stage.
STATISTICS CAN LEAN. THEY NEVER DECIDE.
The rule. Statistics may influence branch preference and directional weighting, but never override structural qualification, reaction confirmation, risk controls, or execution authority. The level is still resolved only by what the market does, and the first chain to finish still wins.
What goes in. Every CONFIRM and every FINAL ENTRY gets one CONTEXT line under its plan, worked out only from what was known when that stage fired. Two parts. The trade-type preference asks whether the session favours breakouts or reversals — from the Previous Month 50% zone (two cards on) and from NS Market Regime, which abstains today because it publishes volatility levels, not trend or range. The directional votes: the Previous Month 50% zone; a setup heading toward the nearest Midnight level the day has not touched votes for, one heading away from the only untouched level votes against; the opening-range profile (next card) votes after its range breaks; the London levels and the Reactive NQ Grid abstain, because they publish no direction. Each part is the average of the votes it received, and the two are averaged with equal starting weights into a score from −1 to +1.
What you see. CONTEXT A · NYAM · S +0.83 · PM50 ↑ EXPANSION (P 1.12) ✓ · TBR continuation 68% ✓ · MRS abstain · ALN abstain · RNQG abstain, or in London CONTEXT A1 · LONDON · S +1.00 · PM50 ↑ EXPANSION (P 1.12) ✓ (single source). A: the trade type and the direction both lean this way, backed by at least two separate sources. A1: the same lean, but from one source only. C and C1: the mirror, against the alert. B: anything in between. B·NO_CTX: no source had current data. When price first meets an armed pair, the DECIDING line can add context prefers BUYLOW (A +0.75) over BREAKDN (C −0.75). The dashboard shows the outcome by context grade; below 30 trades it reads SMALL SAMPLE and is reporting only.
Under the hood1 technical noteThe Opening-Range Profile
WHICH WAY THE OPEN BROKE, AND HOW OFTEN THAT HELD.
Read, not rebuilt. Time Based Ranges - The Daily Profiler by Trades-Dont-Lie is a protected study. VCS reads only what it draws on its background chart — never your chart, never during Bar Replay — at each five-minute close of a window instance on weekdays, for NQ1! and GC1!. The default NY AM instance watches the 09:00–09:30 ET range and votes from 09:30 to 10:30, when the NY AM window closes; an alternative 09:30–10:00 instance is there if you prefer it, and a NY PM instance (13:00–13:30, voting 13:30–15:00) joins once you seed it. It reads the direction of the first breakout and the study's continuation and reversal percentages for that breakout from its own history (2006–2026). Each instance has its own seed and its own settings check, and it votes only inside its own window.
One vote, or none. After the first breakout, if continuation is 60% or more, a breakout alert in the breakout direction gets a vote for it and a reversal back against the breakout gets a vote against; if reversal is 60% or more, the opposite. Otherwise TBR abstains, and every other alert gets no vote. An abstain always says why: the range is still forming, no breakout yet, outside the window, the study is not loaded or its settings differ, or the reading is stale. If the study itself changes, VCS stops reading it until the reader is updated, and the dashboard says so.
What you see. Under the CONTEXT line: TBR NQ NYAM 09:00–09:30 · broke UP 09:35 · continuation 73% · WITH ✓ · MFE med +17 bps · heat med 12 bps, or TBR abstain · NO_BREAKOUT_YET.
FOUR WINDOWS, ONE CLOCK — AND THE 8:30 RELEASE IN PLAIN SIGHT.
The windows. Asia 7:30–10:30 PM · London 2:00–5:00 AM · NY AM 8:00–10:30 AM · NY PM 1:30–3:00 PM, New York time, daylight-saving safe — lined up with the London and New York killzones and the Daily Profiler sessions. NY AM opens at 8:00 so the 8:30 release and the first hour of the Midnight levels sit inside it. One definition drives everything: campaign names, how long an armed setup stays valid, the deploy rule, the grading cutoffs and the research buckets.
The 8:30 tag. A CONFIRM or FINAL ENTRY that fires on a weekday between 8:25 and 8:45 AM carries ⚠ 8:30 NEWS WINDOW — CPI, jobs and claims days move fast both ways and spreads widen. If you list the day's release in the news calendar, the tag names it: ⚠ 8:30 NEWS · CPI. VCS never guesses a release date. The tag is a label: it never blocks, delays or hides an alert.
Grading the way you trade. Every stage outcome settles at its own window's close-out if neither the stop nor the target was hit first: an Asia entry by 2:00 AM, a London entry by 8:00 AM, New York entries by 4:00 PM. The cut-offs are starting values in one file — change them to match how you actually exit. Every grade records which rule it used, and older grades are never rewritten.
Under the hood4 technical notesThe Previous Month 50% · Continuous Structural Reassessment · Clear State. Preserved Evidence. · Alert Rotation Capture
WHERE LAST MONTH'S MIDDLE SITS DECIDES WHAT KIND OF DAY IT IS.
The levels. Last completed month's high, low and midpoint — PMH, PML and the PM50 — from VCS's own daily bars, with the month counted the way each market trades (futures roll at 6:00 PM, currencies at 5:00 PM). P measures where price sits in that range: 0 at the low, 1 at the high.
Five zones. Above the high is EXPANSION UP, below the low EXPANSION DOWN — one-sided, trend days. Around the middle (P 0.40–0.60) is BALANCE — two-sided days. In between, LEAN UP and LEAN DOWN. A zone changes only when a fifteen-minute candle closes inside the new zone, so a spike through a line changes nothing.
Two votes, every window. Trade type: in EXPANSION a breakout gets a vote for and a reversal a vote against; in BALANCE the opposite; LEAN says nothing. Direction: the up zones favour longs, the down zones favour shorts, BALANCE says nothing. That is why Asia and London — where nothing else votes — now read A1, C1 or B, and why New York can reach a full A once the Midnight levels or the opening-range profile agree. An alert level within a quarter of the average daily range (0.25 daily ATR) of PM50, PMH or PML is tagged PM50 confluence.
Honest limits. The idea comes from the Daily Profiler, whose exact 70/30 vs 50/50 rule is taught in its bootcamp; VCS uses a public approximation, and every threshold is an unvalidated starting value. A month VCS cannot fully see, a contract roll that distorts the range, or a missing price means PM50 abstains and says why.
CURRENT EVIDENCE. VALID SETUPS. VERIFIED ALERTS.
An explicit arm with complete causal LOCATION evidence establishes a bounded managed scope: the exact instrument/feed, timeframe, strategy options, branch eligibility, expiry and session. The existing retest resident can then request fresh structural evaluation through the arm owner after lifecycle work. It reuses the current verdict, geometry, ALE and transaction owners; it does not rerun EXECUTE on a timer. Evaluation follows completed source bars with bounded retry/backoff and at most one 35-second acquisition per pass. WATCH keeps its existing intrabar behavior. A missing feed, wrong chart or incomplete read is reported as unavailable.
Monitoring has its own control. Use node vcs_arm.mjs --monitor=OFF, --monitor=OBSERVE or --monitor=PRODUCTION. OFF is the default; OBSERVE reports without changing setup or alert authority; PRODUCTION permits eligible structural changes inside the managed scope. Add --symbol=GC1 to control that existing scope. Global OFF clears existing symbol enables. A disabled symbol remains disabled across restart and global reenable; exact symbol-scoped PRODUCTION may reauthorize an unexpired scope. Lifetime ends at the earlier of original TTL and the contiguous campaign segment boundary, so OFFWINDOW cannot resume after another session. Legacy plans are not adopted, and a one-shot near override grants no continuing override. Use --monitor-status to inspect the state.
Preserve what remains valid. A distant branch survives while its canonical structure remains valid. A separate new opportunity must satisfy the existing coexistence and opposing-branch rules. Authoritative invalidation or causal supersession retires only the affected authority, even when no replacement is eligible. Missing or stale evidence does not prove validity or invalidity. Fired, retest and READY episodes keep their existing lifecycle rules. New E2 identities follow causal market evidence and tick-normalized structural anchors; ordinary price/ATR noise, invocation time and notification changes do not create new opportunities. Historical E1 identities remain readable, and terminal epochs cannot restart without a genuinely new causal occurrence.
Replacement and restoration are separate permissions. Independent node vcs_arm.mjs --condition-repair=OFF or --condition-repair=PRODUCTION controls installed-condition repair; monitoring enablement does not grant it. AUTOREARM/1 continues to govern missing alerts; stopped or expired alerts are not missing. Actual-condition checks compare feed, fixed level, crossing operator, timeframe, frequency, condition structure and active/expiry semantics. Dynamic expressions are checked through inspectable configuration; unknown properties remain UNVERIFIED. A matching tag or successful create response does not prove the installed condition. Authorized mutations recheck scope and current state under the shared transaction lock, persist retirement, preserve attributable fire evidence and read back the result. Partial cloud failure cannot revive obsolete authority.
Configuration never counts as a crossing. Moving or creating an alert cannot manufacture EARLY, CONFIRM or READY. Historical discovery evidence is distinct from a qualifying live event after installation; old-epoch events cannot authorize the new one. Gate 6, proof alternatives, no-chase and risk checks remain unchanged. A definition marked Active is labelled ARMED DEFINITION and is still waiting for its condition; received CONFIRM is a separate canonical event. Status distinguishes loaded-code health, acquisition, market-evidence time, evaluation, desired-plan commit, actual-condition verification and lifecycle progress. No live-transition claim follows from a healthy PID or an offline test; release and activation evidence is recorded in the current-state ledger.
Current limits are explicit. Every production arm requires durable causal registration, monitor PRODUCTION and a healthy loaded reassessment resident. Current-day anchors can remain verifiable when prior-day levels are unavailable. Fresh source-matched barset prices and canonical distance are checked before and after creation. Reversals can wait for an intentional future retrace/reclaim; price side alone proves no past crossing. WATCH approach direction is independent. Scope recovery uses --reconcile-managed; --monitor-status reports owned/orphaned coverage, evidence freshness, failures and overdue work. E2 server handoff and webhooks remain LOCAL_DEPENDENCY_BLOCKED because that server contract lacks installation/configuration lineage. Local canonical lifecycle processing remains available. TradingView cloud notifications while the PC is off do not provide structural reassessment or E2 server READY authority.
Read the actual branch. An armed alert definition is waiting, provider Active is delivery status, and a received event needs exact epoch and branch attribution. The dashboard shows that branch's area, stop and targets, latest event, current state, next proof and reason. Without live inspection, installed conditions are UNVERIFIED and current market evidence is UNAVAILABLE. READY requires canonical proof. An ARMED setup is never relabelled READY merely for distribution.
Scores are evidence, not win probabilities. BPE, regime, range-mode and failed-auction outputs are heuristic scores or classification confidence, UNCALIBRATED. Equal starting weights are a declared baseline prior. Missing evidence is unavailable, never zero or a 50/50 forecast. Raw diagnostic history and all weights, thresholds and authority remain unchanged.
Reduce work without removing evidence. One command now reuses its identical Daily read; arm, retest and post-install freshness checks remain. VCS_ARC_DISABLED=1 skips optional ARC-only work while canonical lifecycle continues. Keep the required QT and IFVG studies: optional graphics and generic study values may still be acquired. Public QT on-chart display was already off in the inspected layout, so no live study was hidden.
MEASURE EACH LEG WITHOUT MERGING THE ALERTS.
Every BUYLOW, SELLHIGH, BREAKUP and BREAKDN fire begins its own durable ARC excursion record. EARLYTHREAD/1 also represents a qualifying opposing EARLY as a linked thesis rotation: the prior directional thesis failed or yielded to an opposing EARLY thesis. ARC remains measurement-only; rotation never fabricates invalidation, deletes a surviving branch, or grants capital authority.
ARC reports R only after the same fired branch is already READY and its canonical entry/stop frame supplies a legitimate denominator. Before then, raw price movement is retained and R is stated as unavailable — ATR is never substituted and a stop is never invented. The 0.25R, 0.50R, 0.75R and 1.00R marks are measurements for later validation, not scalp targets or permission to trade.
THE FIRE IS A LOCATION PING. THE ENTRY REQUIRES COMPLETED PROOF.
The bar that finally closes through a level is often the extension top. Three hard rules on every fire:
Under the hood2 technical notesNative Reversal Confirmation · The Alert, Versioned
STRUCTURE ARMS. CONFIRMATION QUALIFIES.
Normal reversal routes require accepted, direction-aligned M1 IFVG evidence with verified feed, settings, chronology, age and coverage. An unavailable source keeps a structurally valid setup pending. Breakouts use their existing structural proof classes.
For EARLY_DIRECTION routes that require WATCH, the source must satisfy that exact setup epoch's persisted WATCH boundary. Prior-epoch evidence cannot qualify a new episode. Qualified CONFIRM is a canonical structural transition, not an indicator label.
Evidence expires after 20 minutes and needs fresh native observations within 90 seconds. Opposite-edge close invalidation is permanent. Later Gate-6 proof and current price must both satisfy the 2.5-ATR no-chase rule. Only the exact Range Mode ARM/eligible full-retest cohort is source-optional; qualified CONFIRM and reaction hold require IFVG.
On GC, NQ, GBPUSD and USDJPY (REACTPROOF/1) the IFVG counts only after a qualified reaction at the armed level. The first choice is a Liquidity Sweep of a pivot at the level; otherwise a native reclaim qualifies. The IFVG must arrive after the touch and within two completed M5 candles. No reaction means no CONFIRM, but a missing sweep alone never refuses.
SAME FOUR BRANCHES. SHORTER MESSAGE. RICHER RECORD.
The August 14 refactor changed what an alert carries, and nothing about what it is. The four families — BUYLOW, SELLHIGH, BREAKUP, BREAKDN — are untouched, and no fifth family exists because new layers do. What changed:
THE SENSOR LISTENS.
VCS QUALIFIES.
Location alert → REACT WATCH → qualified IFVG reaction → REACT LONG/SHORT → Gate 6 → READY. A raw source event is never a trade signal.
Under the hood3 technical notesIFVG Cloud Sensor — Event Driven · Check What Is Actually Running · Local Hosting Limits
IFVG CLOUD SENSOR — EVENT DRIVEN
Why is it quiet? A normal candle close is not an IFVG event. HEALTHY_IDLE means the verified sensor and local receiver are healthy while waiting. Silence does not refresh source proof; the 90-second evidence limit still applies.
Source cloud alert: a persistent private M1 IFVG machine sensor sends confirmed evidence by webhook, with Notify in App and Show toast notification ON. Email and sound remain OFF. Only new committed accepted IFVG events send; no new event means silence. Several events on one close share one bounded batch. It can listen when no branch exists.
Branch alert: an existing VCS location and exact setup epoch, with its own validity and human delivery settings.
REACT: WATCH means an owned location is being monitored; LONG/SHORT means the canonical owner accepted the qualifying reaction.
READY: later canonical Gate-6 proof. VCS places no orders.
CHECK WHAT IS ACTUALLY RUNNING
September 22 alert coverage repair: EXECUTE now evaluates native M5 and M15 independently. An empty fresh scan can remain under ordinary native observation until the campaign ends. Expired unfired observations release prospective slots; active and fired branches keep their ownership. Per-source candidate and installation diagnostics expose refusals. See ALERT_COVERAGE_RELEASE.md. Qualification and execution safeguards remain unchanged; local observation requires the awake connected PC.
September 22 source repair: GC, NQ, GBPUSD, USDJPY and BTCUSD now share an exact timestamp/OHLC comparison. Native and cloud volume differences no longer block identical price evidence. Real price revisions still block confirmation, with the conflicting fields identified. Original evidence, detector rules and execution safeguards are preserved. See SOURCE_BAR_REPAIR_RELEASE.md for verification.
Delivery repair: futures symbol formatting now produces valid JSON. A missing older minute cannot reject a new event with complete subsequent history; a gap within required event history still blocks it. SOURCE_REJECTED identifies a receiver refusal. Read the REACT212_RELEASE.md delivery repair report for current verified receipts.
September 20 fleet deployment: all 5 required M1 sources are installed and ACTIVE: COINBASE:BTCUSD, COMEX:GC1!, CME_MINI:NQ1!, OANDA:GBPUSD and OANDA:USDJPY. Notify in App and Show toast notification are ON for every sensor. Creation, renewal and resident reconciliation enforce these settings. The event-only hotfix replaces the old immutable snapshots. See the REACT/2.1 release report for current alert IDs, native parity and live observation evidence. All five have observed November 17, 2026 expirations; supervision warns seven days before expiry.
The release report separately records private source compilation, actual Alert Manager IDs and active status, expiration, webhook receipt and human delivery. Compilation alone does not create an alert. An empty Alert Manager fails required sensor deployment.
Cloud observer lifetime, branch lifetime and evidence freshness stay separate. IFVG retains its 20-minute event cap and 90-second observation/native-close limit. Local and cloud observations share event identity.
LOCAL HOSTING LIMITS
The operator selected the existing local host. The TradingView sensor runs in the cloud; VCS qualification and the tunnel require the awake connected PC. PC-off qualification is blocked. Qualified REACT mobile transport is unavailable. Raw IFVG notifications are enabled separately; they do not prove a qualified reaction or device receipt.
Breakouts keep their native continuation proofs. REACT/1 is retained below as historical implementation context; its source/deployment receipts are not rewritten. REACT remains statistically UNVALIDATED.
LOCATION FIRST.
REACTION NEXT.
Your existing alert identifies the location. REACT presents qualifying evidence at that location. The existing readiness checks decide whether the setup is complete.
Under the hood2 technical notesOne Proof. One First Reaction. · Three Clocks, Three Different Jobs.
ONE PROOF. ONE FIRST REACTION.
BUYLOW and SELLHIGH use eligible bullish and bearish M1 IFVG respectively at qualified CONFIRM. On the four REACTPROOF markets that IFVG is paired with the reaction at the armed level, shown as one line (SWEEP M5 … → … · reclaim … or RECLAIM M5 … · close …). The local message says REACT LONG or REACT SHORT and remains NOT READY until the required later Gate-6 proof. M15 provides structural context; recorded branch intervals and structural stops stay authoritative.
BREAKUP and BREAKDN retain their native full, shallow and no-retest continuation routes. Their REACT context enriches READY.
THREE CLOCKS, THREE DIFFERENT JOBS.
Observer: persists across setups on the enrolled current chart. Branch: retains its canonical epoch, deadlines and invalidation. Evidence: IFVG still expires after 20 minutes, with 90-second observation and native-close freshness. A persistent observer never makes an old signal valid again.
New gaps cannot reset the first-proof claim. READY absorbs an unsent CONFIRM update; stale and terminal notices are suppressed. Restart retains claims. An uncertain desktop send is reported as uncertain, not automatically repeated.
WHAT REACHES YOUR PHONE?
Existing TradingView WATCH, EARLY and CONFIRM alerts keep their app and toast notifications. Qualified runtime REACT is currently a local desktop capability: qualified REACT mobile delivery is unavailable. Desktop transport requests are not proof of human receipt. Local processing requires an awake connected computer.
The native observer has no cloud expiration or renewal. It never changes the active symbol; other enrolled symbols can become stale. Healthy silence is different from broken coverage. Market closure is reported only for verified calendar coverage.
Under the hood1 technical noteCheck Or Pause Observation.
CHECK OR PAUSE OBSERVATION.
node vcs_confirmation_read.mjs --status shows source identity, coverage and health. Use --pause=<exact-observer-id>, --resume=<exact-observer-id> or --remove=<exact-observer-id> for that observer. A manual pause survives restart and new settings generations.
WAITING has no reaction proof yet. REACT has recorded proof. READY is proof, not an order. Terminal setups cannot revive. Engineering verification does not establish a statistical edge.
QUARTERLY
THEORY.
VCS reads price and volume — Quarterly Theory adds time. Every cycle splits into four quarters, and each quarter has a job: it tells you when the range builds, when the sweep hits, and when the expansion runs. This is the Timing in "Timing, Location, Acceptance."
Markets move in repeating time cycles. Quarterly Theory splits every session, day and week into four quarters — accumulation (the build-up), manipulation (the fake-out), expansion (the real move) and continuation — so you know where in the cycle you are. It is a map of time, never an entry signal.
The official NY AM alert window opens at 8:00 AM; Quarterly Theory’s own morning states start at 9:00 AM. They are two different clocks.
SAME PATTERN, EVERY SCALE.
THE QUARTERS MAP TO YOUR SETUP.
STACK THE CLOCK WITH THE CHART.
Run it in auto mode showing two layers from your chart timeframe — your cycle plus the one above it for context. Then stack the alignments:
THE CLOCK SAYS THE QUARTER. PRICE CONFIRMS IT.
The bot now pairs the pure clock read with an hourly price-aware read: did this hour sweep the prior hour's high and reverse (a C2 manipulation candle), or take it with the trend (C3 continuation)? Is the 3-hour structure a LINE (keep going) or an APEX (expect the turn)? When the clock and price disagree, price is the tie-breaker.
Under the hood1 technical noteThe Morning Window
09:00–10:30 ET IS THE PRIMARY WINDOW — AND CAUTION IS A CAP, NEVER A VETO.
Session timing used to live in three overlapping rules that could stack against the same trade. It now lives in one place, as four named states: pre-open reduced risk (before the cash open) → elevated caution (the open itself) → full authority (the settled morning, through 10:30) → outside the primary window (everything after). One owner, one deduction.
tradingview.com/script/nGqjV24C-Quarterly-Theory-Cycles-Oracle
VCS BREAKOUT®
When price compresses tightly against a defensible structural boundary, the range is primed to break. VCS BREAKOUT trades the expansion out of the range rather than the fade at its boundary — and its authorization is structural: the M15 body close and the retest.
When price coils tight against a real level, it is primed to break. VCS BREAKOUT rides that break — after proof.
TIGHTER COMPRESSION, BIGGER BREAK.
Run the compression read top-down (D → H4 → H1 → M15) showing progressive contraction. The tighter the compression — lower compPctile, coil3 firing — the higher the breakout potential. Compression building directly against a prior-day or session extreme is the highest-energy shape: stored pressure against a real wall.
FIND THE TRIGGER LEVEL.
The breakout trigger is the nearest defensible structural level beyond the coil — the prior day's high or low, the current day's extreme, a session extreme — with the same guard the fade uses: a level closer than 0.25× ATR is coil noise, not a boundary.
CLOSE THROUGH IT — DON'T WICK.
Price must BODY-close beyond the boundary on M15 — not just wick through. A wick through the level without a close is still range: do not enter. Acceptance = at least one full M15 candle body closing beyond the structural boundary. The alert enforces this server-side — a wick physically cannot fire it.
CONFIRM THE STRUCTURE.
Use H1 as your primary indication timeframe; if H1 structure is unclear, use H4 for context. Read the indication level off price itself — the swing that produced the break, and the structural level it broke from. That level is the reference the correction entry is measured against.
ENTER THE EXPANSION — ON THE RETEST.
| VCS RANGE | VCS BREAKOUT | |
|---|---|---|
| Core Strategy | Mean Reversion · Fade the Boundaries | Momentum Expansion · Ride the Break |
| Market Condition | Ranging / Choppy / Consolidation | Tight Compression into Expansion |
| Entry Style | Rejection at Range High/Low + Reversion to the Center | M15 Body Close Beyond the Structural Boundary, entered on the retest-hold |
| Primary Target | Range Center | Next Structural Level (prior-day / session extremes) |
| Base Profit Objective | +0.50R first — extend toward the center, then the far boundary, when the trade supports it | +0.50R first — extend toward the next structural level when the trade supports it |
| Stop Loss | Just beyond Range Boundary | Just back inside the Broken Level |
| R:R Determination | Gate 6 entry → structural invalidation → range center / valid reversion objective → natural R:R | Gate 6 entry → breakout invalidation → next structural expansion objective → natural R:R |
| Key Signal | Boundary Compression + Bar-Level Rejection | Tight Compression + Structural Boundary + Clean Close |
| Best For | Choppy & Ranging Markets | High-Momentum Breakouts |
Replay evidence remains research, not live authority, and the current proof-class cohorts are not yet validated edges. Both families stay independently armed when valid, and neither suppresses the other; a breakout must earn its own structural confirmation and completed Gate-6 proof, never borrowed IFVG authority.
THE BOT.
Claude Code reads your TradingView charts over the MCP bridge, runs the full VCS pipeline as numbers, not screenshots, arms the alerts, and leaves graded setups waiting for you. The cloud holds the branch and ladder alerts — your machine can be off when those fire. You stay the pilot.
The bot is Claude Code connected to TradingView. It reads your charts as numbers, runs every VCS rule, arms the alerts and follows each setup to its proof. It never places an order.
YOUR SYSTEM, RUNNING ITSELF.
The TradingView MCP connects the agent to TradingView Desktop through the Chrome DevTools debug port. One canonical launcher owns Windows/macOS startup, verifies the CDP endpoint and TradingView process, and reuses a healthy warm process instead of spawning duplicates. VCS routes one product, acquires one immutable sweep, grades it once, and arms through the canonical path.
This is a fusion, not a replacement. The MCP gives Claude eyes on the market; your framework gives it judgment; the alert ladder gives it delivery. What comes out is a cockpit — setups surfaced, graded, and waiting — while you confirm and pull the trigger.
Under the hood2 technical notesFast Without Changing The Trade · Evidence Lanes Stay Separate
ONE OWNER. ONE SWEEP. ONE GRADE.
Speed work changed coordination, startup, chart readiness, acquisition and reuse — never the strategy. One conductor delegates to the existing owners; a compact generated kernel carries only live operating law; internal ticker/timeframe identity prevents false ten-second legend-label waits; and each feed is selected once. FULL and FAST_REFRESH are still chosen automatically per campaign: the first valid campaign run is FULL, later valid same-campaign runs refresh M15/M5, and stale, corrupt, mismatched or uncertain state fails closed or returns to FULL.
RELATED MARKETS ARE NOT INTERCHANGEABLE AUTHORITY.
NQ1! and GC1! are the authoritative analysis markets for their futures families; MNQ and MGC are traded products, never substitute analysis sources. NAS100 and XAUUSD remain separate CFD lanes. BTCUSD is an explicit crypto-spot analysis-only command. Related feeds may provide declared context, but CFD evidence cannot authorize a futures branch and futures evidence cannot authorize a CFD branch.
TWO PRINCIPAL FUTURES MARKETS. NEITHER IS SECOND.
The futures workflow runs on two primary VCS analysis markets: NQ1! — Nasdaq-100 futures — PRIMARY, and GC1! — Gold futures — PRIMARY. Each anchors its whole family (the micros MNQ and MGC are execution vehicles that draw analysis and evidence from the full-size chart, never separate identities). Feature support is documented per market and is never a statement of priority: both primary families carry the PSY/1 psychological grid (Ch 03) — Gold on the source's own verified defaults, Nasdaq on an explicit user-approved port whose geometry is honestly stamped UNVALIDATED until separately proven. When Nasdaq lacked the layer that was scope, not demotion; now that it has it, that is feature support, not promotion. NQ1! keeps its full analysis, alert and gate stack.
AN ALERT PRICE CAN'T REACH IS NOT A PLAN.
Before anything arms, every level is checked for reachability: a branch beyond 2.5× the current ATR is blocked — price will not get there before the next look. This is the best-measured filter in the stack: blocked branches fire three times less often and lose money when they do. When every edge is out of reach, the run stands down and names the structure worth watching instead — a legitimate outcome, never a failure.
Under the hood1 technical noteContext Engines
THE BACKEND READS THE ROOM — AND STAYS IN ITS LANE.
Behind the gates, a set of context engines classifies the tape: a regime engine (trending / range / expansion / transition, plus a volatility state), an auction-context layer (a nine-scenario daily hypothesis with the discipline to say NO CALL), a branch-evidence lean for every armed boundary, and an attention ranking across everything armed. Every one of them is advisory: they label and log, and none of them can move a grade, a level, a trigger, or an arming decision until it measures positive. That is by design — see Chapter 09. Each is currently stamped recalibration required, and statistics from different input versions are never pooled.
SEE EVERYTHING ARMED, AT A GLANCE.
A read-only dashboard renders every armed branch: the evidence lean per boundary, the lifecycle state (armed → at level → triggered → awaiting retest → ready), the exact next required event, the trade plan, and data freshness. It changes nothing — it is a window, not a control panel. A watchdog heartbeats the background watchers every five minutes and restarts them if they die.
What it costs to keep running once you are configured. Check the current prices and capabilities of the plans you choose.
Required. Powers the charts, indicators, and cloud alerts. The ladder shares the 100-alert budget: optional WATCH Scouts drop first, while the EARLY heads-up and the CONFIRM entry alert are protected. If those cannot fit, the arm refuses before mutation. The CME real-time add-on remains mandatory when trading those contracts.
Required after setup. The $200 Max plan is for the initial setup month only — the full package analysis, installation, and first verification pass are sized for Max headroom. Once installed, downgrade to Pro: it runs the daily agent that reads your framework, drives the MCP, grades the setups, and arms the alerts.
Running cost after setup: $60/month, all in — $40 TradingView (once the trial ends) + $20 Claude Pro. Prices as of August 16, 2026 and subject to change.
A MID-RANGE LAPTOP RUNS THIS.
You do not need an expensive gaming rig or a maxed-out Mac. The job is modest: TradingView Desktop, Claude Code in the terminal, and the MCP bridging them. Put your money into your trading account, not silicon you will not use.
CONNECT CLAUDE TO TRADINGVIEW.
Open Claude Code and paste the install prompt. It clones the TradesDontLie TradingView MCP, installs dependencies, registers the server in your MCP config, and launches TradingView with the debug port so Claude can read your charts live.
LET THE BOT RUN UNINTERRUPTED.
Before launching, bypass permissions so the bot runs fully automated in the background without stopping to ask for approval on each action:
POINT IT AT THE VCS BOT FOLDER.
You don't drag a document in anymore — the portable VCS BOT folder (or VCS BOT (Windows + Mac), the cross-platform build) is the system: the four bot prompts, this guide, the grading rules, the engine, the test battery, and the installer, all versioned together. Launch Claude Code from inside that folder and enter:
START IT EACH TIME YOU TRADE.
Open Claude, go to the Code section, start a New Session, and give it the EXECUTE command for the market you trade. This is the one command you run at the start of every trading day — EXECUTE plus your asset:
Compile and save the attributed IFVG companion privately, then verify exact saved source, settings and native output. Use standard M1 candles. Source confirmation for MNQ comes from NQ; MGC comes from GC. CFD feeds remain separate.
Keep TradingView and the retained VCS watchers connected. A fresh loaded-code heartbeat is required; a process identifier alone does not prove healthy monitoring. Missing evidence remains unavailable and never bypasses execution checks. The source reader needs no cloud study alerts, public receiver or tunnel.
WATCH, EARLY and CONFIRM price alerts retain their existing notifications and ownership. VCS places no orders.
TradingView and the retained VCS watchers must be awake and connected for local confirmation and READY processing. Their loaded-code health must be current. No execution-critical model call is required.
Check loaded watcher health before the session. An explicitly armed, unexpired managed scope can be reassessed by the existing resident when its independent monitoring control permits it.
The normal analysis output records qualified branch plans. Resident structural work stays within explicitly authorized managed scopes; an armed definition still waits for its condition and canonical proof. Your job is to confirm the entry on the charts and place the trade through your broker as usual.
Once the branches are armed, TradingView's cloud holds and fires them — your PC can be off. The fire is re-checked live when you act: the alert carries placement-time analysis, and the market may have moved on.
Discretion is always advised. Always buy low and sell high. Do not — I repeat — do not enter blindly.
Your session's time window opens — a simple clock decision, or the scheduled sweep that runs at it. This is the signal to run analysis, nothing more.
The bot pulls the full stack as numbers across Weekly → M5 on both feeds — structural levels, market state, compression, time-of-day — and grades every setup against the framework. About three minutes per instrument.
The Fork box states the lean and the line in the sand. Every valid, reachable edge is armed two-sided with the full alert ladder. Skipped branches are named with reasons — nothing disappears silently.
WATCH prepares the location. EARLY opens the directional thesis as a heads-up. CONFIRM, on that same thesis, is the primary entry alert. A valid opposing EARLY produces a linked ROTATION. None creates an order; completed proof alone may reach ALE READY.
Execute only after canonical ALE READY: RETEST_HOLD, reversal REACTION_HOLD, breakout SHALLOW_RETEST, or breakout NO_RETEST_CONTINUATION as family-appropriate. A completed close back through the level kills the branch; overextension is MISSED — DO NOT CHASE.
LIVE
CONTEXT.
The live sources around the stack — the Nasdaq sessions, the midnight revisit, the volatility regime, the gold reference map, the reaction proof and the location owner — and the integrity rules they share. Each one carries a stated authority it cannot exceed.
These live sources sit around the core rules. Some decide where a setup’s level is; the rest only describe the market.
Under the hood4 technical notesLive NQ Sessions — ALNCTX/1 · Midnight Revisit — MRS/1 · Live H1 Context — NSREGIME/1 · Live Context — NOMADA_CONTEXT/1
ASIA, LONDON, AND WHAT NEW YORK USUALLY DOES NEXT.
WHERE MIDNIGHT TRADED, AND HOW OFTEN THE DAY CAME BACK.
ONE HOURLY READING, SHOWN ON THREE VIEWS.
TWO CONTEXT SOURCES, ZERO NEW AUTHORITY.
THE LEVEL MUST REACT BEFORE THE IFVG COUNTS.
Under the hood3 technical notesLocation Owner — NRA/1 · September 20 Integrity Audit · Source Settings And Coverage
THE NOMADA LEVEL IS WHERE. THE PROOF IS STILL VCS.
SEPTEMBER 20 INTEGRITY AUDIT
The first completed candle after a fire is checked for invalidation before proof. Lifecycle evidence requires standard price bars. Research keeps unknown outcome bases outside the real track record. These repairs preserve the current strategy, alert roles and execution gates.
See the principal-engineer report dated September 20, 2026 for exact offline verification, distribution receipts and remaining live-operation limits. Software checks do not prove a statistical edge or phone delivery.
SOURCE SETTINGS AND COVERAGE
IFVG must match its verified detection preset. A settings mismatch is reported separately from a source-version failure, with actual and expected values. Restore the verified preset and acquire fresh native evidence; normal reads never change settings. The shared repair covers the registered asset set, while closed-market bars remain stale. These checks do not add trading authority or prove a statistical edge.
THE MEASUREMENT
LAYER.
What separates VCS from every discretionary system it resembles: it measures itself, and the measurements have teeth. Layers get demoted. Points get withdrawn. Results that flattered the system get re-run honestly. This chapter is how that works — and why you can trust the numbers the bot quotes.
VCS keeps score on itself, honestly. Every setup freezes its evidence when it is armed, every outcome is graded from VCS’s own price data, and no number is called an edge until it passes the audit standard on enough trades.
FOUR TESTS, THREE SAMPLE FLOORS.
No edge number is trusted until it survives a four-step audit: a holdout split (does it work on data it wasn't fit to?), a multiple-comparisons correction (did we just try enough configs that one looked good?), a bootstrap confidence interval (does the interval exclude zero?), and a walk-forward test (does it hold in time order?). Sample floors: 100 observations minimum, 300 for confidence, 500 to be production-grade. Below the floor, a statistic is reported as unmeasured — never as an edge.
Under the hood1 technical noteThe Fill-Model Reckoning
THE BEST RESULTS WERE THE FILL MODEL. SO THE FILL MODEL WAS FIXED.
A replay audit found the backtester was assuming entries at prices the live rules would never give — and that assumption, not the market, was producing the system's best-looking numbers. The fill model was rebuilt to mirror the live entry discipline (the Gate-6 retest fill), and every result was re-run. Several celebrated edges collapsed to zero. They are gone from this guide.
WHERE THE MEASURED EDGE LIVES.
ADVISORY UNTIL PROVEN. DEMOTED WHEN DISPROVEN.
Every new engine starts log-only: it computes, prints, and records — and changes nothing. Promotion to a live gate or score requires the full audit pass on adequate samples plus explicit sign-off. The reverse also happens, on schedule: a layer that measures negative loses its vote. The two-sided coil of system building: arm both branches, let the evidence pick.
Under the hood4 technical notesThe Learning Loop · Time & The Opening Hour · Prealert — Confirmed Early Reactions · Alert Integrity — Pulse & Setup Outcome
EVERY SETUP FREEZES ITS EVIDENCE. EVERY OUTCOME CLOSES ITSELF.
As of this edition the loop from analysis to evidence runs without anyone grading homework by hand. The moment a setup arms, the system freezes a complete snapshot of the evidence that existed at that moment — market state, clock state, location and its provenance, the auction and price-state context, the confirmation reads, the per-boundary probabilities — into an append-only store that nothing is allowed to rewrite. First write wins, forever: a study run next month sees exactly what the system saw at placement, never a reconstruction flattered by hindsight.
Outcomes close the same way. An hourly worker replays every fired branch against cached bars to a stated result — stop-first when a bar could have hit both, expiry never counted as a loss, a missing data source recorded as a gap and never proxied from a lookalike feed. A daily cycle then sweeps, integrity-checks, censuses the unresolved backlog with its reasons and ages, joins the frozen snapshots to the closed outcomes, and refreshes every research table. Real fills stay a separate population from theoretical replays — the one real track record can never be silently inflated.
THE CLOCK LEAVES FOOTPRINTS. THE SYSTEM NOW MAPS THEM.
Markets carve visible highs and lows into named windows — the Asia session, London, yesterday's extremes, last week's. Traders talk about these levels as if they were magnets, and sometimes they behave like it. As of this edition the system tracks them deterministically: every named window's extreme is computed pre-open from cached bars, stamped with its window and clock version, and then followed — untouched, touched, swept by at least one tick, or swept and reclaimed. No bank orders are claimed. What is claimed is checkable: a price, a time, and what happened next.
The opening hour gets the same treatment. The Initial Balance — the 09:30–10:30 range — is recorded objectively once the hour completes: which final extreme printed first, where the hour closed inside its own range, and the classical extension targets measured from it. A crucial honesty rule follows from the clock itself: the finished Initial Balance is a 10:30 fact. A trade taken at 09:45 never gets to borrow it, in live trading or in research replay — the frozen evidence for that trade holds only what was knowable at 09:45.
CONFIRMED BEFORE THE RETEST. MEASURED, NEVER TRADED.
Sometimes the system is simply right early. A reversal boundary is touched, price pokes beyond the edge, a completed candle rejects back through it, the delivery confirmation prints — and price leaves without ever offering the strict retest the entry rule demands. The final alert lane records nothing, and a correctly-anticipated, fully-confirmed reaction would vanish from the books. As of this edition those reactions are recorded as their own shadow measurement cohort: the completed pre-alert sequence plus a valid current-cycle confirmation makes an observation outcome-eligible; a sequence without the confirmation earns nothing — canonical lifecycle proof remains required. The scope is the range-reversal family only; the breakout families keep their full acceptance-and-retest architecture untouched.
The measurement is built hindsight-proof. The clock starts only when both facts exist — the rejection close and the valid confirmation — and no favorable move from before that instant is ever credited. The shadow entry is the first completed close after eligibility, the stop is the branch's own frozen structural stop, and the first objective is the same +0.50R base target (Ch 02) — a starting objective, never a ceiling. A setup whose final alert later fires is linked to its pre-alert twin through one identity and counted as one setup, never two; if the final alert fired first, the final lane keeps the whole story.
SILENCE IS NEVER AMBIGUOUS. EVERY READY IS MEASURED.
PULSE/1 — the dead-man switch. "No setups right now" and "VCS is not running" used to look the same. Now the alert watcher pings an outside dead-man service every minute — a success only when monitoring is in production, both background watchers are healthy and the source feeds are current, otherwise a failure that names its reason. A PC that is off, asleep or offline cannot ping at all, so the outside service raises the alarm on its own. Every weekday a PRE-FLIGHT at 07:45 ET and a DIGEST at 16:15 ET arrive as one desktop notification each.
OUTCOME/1 — the setup outcome. Every READY is graded from its READY price: did it reach +0.50R first, the structural target, or the stop, or was it marked at its window's settle (Asia 02:00 · London 08:00 · New York 16:00 ET) — in R before and after costs, with the furthest move each way and the minutes it took. It is labelled SETUP OUTCOME (not your fill), it never touches your real trade record, and it is history, never a forecast.
Most retail systems accumulate indicators the way charts accumulate clutter — every new idea is additive, nothing is ever falsified, and the backtest always flatters. VCS is run like an institution: principal reviews on a weekly cadence, defects published against itself, and a standing rule that a data gap can never loosen a gate. When the bot tells you a setup is B+ with a data-gap flag, that honesty is this chapter working.
DESKS &
RESOURCES.
LUCID TRADING
lucidtrading.comGATESFX
gatesfx.comOffshore brokers carry their own risk profile — confirm eligibility and your local regulations before funding any live account.
The MCP server that bridges Claude Code and TradingView.
His newest crash-course playlist — the foundation that made market structure click for me.
Bot prompts and implementation records are maintained in the private distribution. Third-party indicator source remains subject to its supplied attribution and permissions.
CREDITS.
QUARTERLY THEORY.
Author of the Quarterly Theory Cycles [Oracle] study, whose time context is part of VCS.
The private IFVG adaptation derives from the supplied CantoLab-origin IFVG Ultimate+ source. VCS modifications add confirmed machine evidence and independent retention. They do not claim authorship of the original algorithm, author endorsement, a new redistribution licence or a validated edge. Original attribution and private source receipts are preserved.
Creator of the ICC Market Structure & Phase Tracking indicator — a tool that lets beginners understand market structure on a genuinely comprehensive level. Her work and her generosity shaped how this system reads structure. Thank you for the inspiration.
A peer trader and developer whose building and perspective sharpened this system. Grateful for the collaboration, the shared problem-solving, and the steady push to turn raw ideas into working tools.
Found out he was a day trader during my second appointment in his chair — I was two weeks into trading when that coincidence hit. Sometimes the universe puts the right person across from you at the right time. Grateful for the early conversations and the encouragement to keep going.
His ICC framework — Indication, Correction, Continuation — is what first made market structure click for me. It was the training wheels I rode until I had mastered structure well enough to take them off and trade without the indicator. The foundation this whole system grew from.
FAQ.
The questions a real person asks after discovering VCS, answered straight — no hype, no jargon. Read these before you risk a dollar.
No. You can begin with simulation, a prop-firm evaluation, or eventually your own live account — the amount of capital matters far less at the beginning than learning to control risk and execute consistently. A prop evaluation lets you trade large simulated capital for a small fee instead of risking savings; a personal live account can be opened small, but real money means real risk. Current recommended desks and brokers live in Desks & Resources (Ch 10). Never fund an account with money you cannot afford to lose.
Yes. VCS does not require you to sit at a chart all day — the alert ladder exists so the system watches price while you live your life. Learning fits around a normal schedule: study a chapter, replay old sessions, run the sim, review the alerts that fired while you worked, journal what you would have done. Consistency beats hours — a steady routine of focused study is worth more than occasional marathon sessions. One honest caveat: not every work schedule overlaps every market session, so your practice window may not be the session you eventually trade.
Separate two different things: understanding the rules and executing them consistently. Most people understand the framework — ranges, boundaries, confirmation, the retest — much sooner than they develop the discipline and pattern recognition to trade it well. Learning what VCS does is the easy part; learning to wait for it is the skill. There is no honest guaranteed timeline. Think in stages instead: understand → recognize → practice → execute → review, and let each stage take the time it takes.
A complete beginner can learn VCS, but VCS organizes trading concepts — it does not replace the need to understand them. Before the framework will make sense, spend a little time on the basics: what a candlestick shows, long vs short, what an entry, stop loss and take profit are, what risk means, and rough ideas of support, resistance and market structure. None of that is hard, and once you have it, VCS gives those pieces a disciplined home instead of leaving you to improvise.
No expert chart-reading skill is required to begin — VCS is designed to organize the decision process so you are not forming dozens of subjective opinions per session. You will, however, gradually learn to recognize a handful of things by eye: structure, boundaries, rejection, acceptance, retests, and the difference between a ranging market and an expanding one. That recognition builds naturally through replay and review; it is a product of practicing VCS, not a prerequisite for starting it.
No. The system and its engines handle the heavy calculation. What you need is arithmetic you can do on a napkin: how much money a trade risks, what percent of the account that is, how position size relates to stop distance, and what "R" means (one R = the amount you risked; a 2R win made twice what it risked). If you can work those out with a calculator, the math will never be the thing holding you back.
Yes — and you should. Paper trade, run simulated futures or demo forex accounts, use bar replay to walk through old sessions, and review historical VCS setups against what the system called at the time. Execution discipline — waiting for confirmation, honoring the retest, taking the loss when structure says so — should be practiced until it is boring, before any meaningful real money is on the line.
Do not measure month one in profit. Measure it in understanding and discipline: learning the vocabulary, recognizing the setup families, understanding what each alert means, absorbing the retest rule, and internalizing the difference between a setup existing and an entry being earned. Expect to make mistakes in the sim and to journal them — repetition and honest review are the actual products of the first month. P&L is not.
Readiness is behavioral, not a magic statistic. You are close when, in simulation, you can repeatedly: follow alerts without chasing, wait for completed proof — the CONFIRM entry alert or FINAL ENTRY — accept MISSED when price runs beyond the entry window, place stops where structure says, size positions correctly, accept a losing trade without revenge trading, and respect session limits. You are closer when following the rules feels routine — not exciting.
For most beginners, yes — one market at a time keeps the cognitive load sane. In the current VCS futures architecture the two primary analysis markets are the Nasdaq family (NQ/MNQ) and the Gold family (GC/MGC); picking either one and staying there for a while builds familiarity with its pace, volatility, session behavior and contract sizing. Neither is "the better market" — no current measurement says so — the point is depth of familiarity, not asset loyalty. Expand once one market feels routine.
Run analysis, review locations, arm valid setups, then wait. WATCH calls attention to the area; EARLY is the heads-up; CONFIRM is the primary entry alert on the same thesis; FINAL ENTRY is the most filtered option. Enter only on completed proof inside no-chase — you place the trade yourself; VCS places no orders.
Start low, and make confirmation earn size: the weakest setups deserve your minimum risk, stronger fully-confirmed setups may earn more, and a hard maximum caps everything. Two rules are non-negotiable. Confirmation earns size — losses never earn size: never increase risk to win back a loss. And position size adapts to the stop — the stop never adapts to the position size: the stop lives where structure says, then size is computed from your risk budget and that stop distance (in forex, risk budget ÷ stop distance gives lot size; in futures, risk budget ÷ (stop distance × point value) gives contracts). Protecting capital comes before growing it — survival is the only early goal.
No. The bot analyzes, arms the alerts and tracks each setup through to READY, the final proven state. VCS is analysis and alerts — you place every trade yourself.
No. WATCH is optional human attention; EARLY is the heads-up and persists as one thesis through CONFIRM (the primary entry alert) and READY. Durable state survives one-shot delivery and restart. Reconciliation never treats a missing projection as new market evidence.
A prop firm lets you trade large capital while risking only the eval fee, but you follow drawdown rules and split profits. A live account gives you full control and 100% of the upside, but your capital is fully exposed. Many traders build discipline on prop accounts first before going live.
RANGE owns BUYLOW/SELLHIGH rejection; BREAKOUT owns BREAKUP/BREAKDN acceptance. Neither a wick nor the first M1 close through the level (the EARLY heads-up) is an entry; CONFIRM is the primary entry alert. RANGE reaches FINAL ENTRY (READY) through RETEST_HOLD or qualified REACTION_HOLD; BREAKOUT through RETEST_HOLD, SHALLOW_RETEST, or NO_RETEST_CONTINUATION. Family-correct completed proof and no-chase decide.
Start with +0.50R as the primary profit objective. It is the base target, not a hard cap — VCS is designed to capture the favorable movement the setup actually provides rather than forcing every trade to reach an oversized target. Once price gives you that first move, taking the full profit is valid, securing a partial is valid, and holding is valid too: if structure, momentum, confirmation and available range continue to support the trade, you can hold for +0.75R, +1R, +1.25R or a larger structural objective. One rule never bends: the stop is never widened to serve the target — risk stays structurally defined. The principle is simple: secure the move first, then let exceptional trades earn the right to run.
The GC and NQ session figures at the front of this guide come from a historical replay of the last 12 months at a +0.5R target — a record of the past, not a forecast. The current IFVG-confirmed cohort is UNVALIDATED. Track prospective, cost-aware outcomes before making performance claims.
Be realistic. Most retail traders lose money, and the ones who last treat it as a business with a long apprenticeship — not a get-rich-quick scheme. The edge is process and discipline, not predicting the market. It is possible, but it takes months to years of screen time and journaling. Anyone promising fast riches is selling you something.
No. Keep your income while you learn. Trade small, prove consistency across many months, and let the results — not the excitement — make that decision for you down the line.
WATCH: price is near an armed level — look. EARLY: a heads-up that VCS now favours a direction — get ready, not an entry yet. CONFIRM: the confirmation candle has closed — the primary entry alert. FINAL ENTRY: the Gate 6 proof is complete — the most filtered entry. Each stage prints its own plan with the entry, the structural stop and +0.50R; NO_PLAN means price is already past the stop or the target.
A, A1, B, C1 or C says how much the background context — last month’s midpoint, the Midnight levels and the New York opening range — agrees with the alert. A and A1 agree (two or more sources, or one); C and C1 disagree; B is mixed; B·NO_CTX means no source had current data. It is a label, not a prediction: it never blocks or changes an alert, and its weights are unvalidated.
Asia 7:30–10:30 PM, London 2:00–5:00 AM, NY AM 8:00–10:30 AM and NY PM 1:30–3:00 PM, New York time. An alert that fires on a weekday between 8:25 and 8:45 AM carries ⚠ 8:30 NEWS WINDOW, because releases such as CPI or the jobs report can move price hard both ways. The tag never blocks or delays an alert — it tells you to be careful.
EVERY TERM,
IN PLAIN WORDS.
The words you will meet in alerts, on the dashboard and in this guide — one line each. Press / to see where each one is used.
ABOUT US.
MEET THE FOUNDER.
Layne Capital is the home of VCS — Volume. Confluence. Statistics. — and of this guide. It was founded by Anthony Dario Layne, who owns the system, designed its rules and develops it. The standard is simple: the market proves it before VCS calls it.
ANTHONY
DARIO LAYNE.
Anthony was born on July 14, 1999, in Bedford-Stuyvesant, Brooklyn, New York, to West Indian parents. His mother is from Barbados, and his father is from Trinidad and Tobago.
His interest in technology began at the age of seven. He brings fifteen years of experience as a graphic designer and video editor, along with three years of experience in music engineering.
Anthony began learning to trade in September 2025, and in May 2026 he began developing his own trading system: VCS.
- BORN
- July 14, 1999 · Bedford-Stuyvesant, Brooklyn, NY
- ROOTS
- West Indian · Barbados & Trinidad and Tobago
- BUILDS
- VCS — Volume. Confluence. Statistics.
AN EDIT LIVES ON TIMING. SO DOES A MIX. SO DOES A TRADE.
Cutting video, designing graphics and engineering music are crafts of timing, structure and rhythm. VCS reads a chart the same way: timing, location, acceptance.
Market structure first. The traders who helped along the way are thanked in Credits.
Eight months in, his rules became a system of his own: VCS.
This guide was first published on June 20, 2026. You are reading its third edition.
An AI trading algorithm that finds the level, watches it and proves the reaction. VCS never places a trade — you make the final call.
PROOF FIRST. YOU MAKE THE FINAL CALL.
CONTACT.
LET’S TALK.
The quickest way to reach Layne Capital is a direct message on X or Instagram. Follow along for VCS updates and every new edition of this guide.
THESE THREE ACCOUNTS ARE THE ONLY OFFICIAL ONES.
Layne Capital will never ask for your password, your trading account login or money to trade for you — and VCS never places a trade on anyone’s behalf. If an account that is not listed on this page messages you claiming to be us, it is not us.
VOLUME. CONFLUENCE. STATISTICS.
@laynecapital ON X
@layne.capital ON INSTAGRAM
@anthonyjustanthony ON INSTAGRAM
BUY LOW AND SELL HIGH.
DISCLAIMER: "VCS RANGE" is an educational resource created by Anthony D. Layne and is not affiliated with, endorsed by, or sponsored by TradesDontLie, TradesBySci, or any of their owners, brands, or affiliates. "TradesDontLie", "TradesBySci" and related marks are the property of their respective owners and are referenced here only for descriptive, nominative purposes. This work is inspired by publicly discussed concepts of volume, support/resistance and swing high/low trading (often called "ICC"), but it presents original explanations, examples, and methods. It does not reproduce proprietary course material, private indicators, or paid content belonging to TradesBySci or others. Third-party indicators, brokers, prop firms, and software are referenced for convenience and are not endorsements; each is the property of its respective owner and carries its own terms and risks. Trading involves substantial risk of loss and is not suitable for every investor. Past performance does not guarantee future results. Backtested and replayed results are hypothetical, are derived from historical data, and do not represent actual trading. Nothing in this guide is financial, investment, legal, or tax advice — do your own due diligence and consult a licensed professional. Product availability and pricing references retain their original observation dates and are subject to change; this documentation update does not reverify vendor prices.