PSRC Meridian
- Sandra Wakefield
- Jul 23
- 8 min read
Updated: Jul 25
Access the Whitepaper
Executive Summary
PSRC MERIDIAN is a three-layer institutional confluence framework built for TradingView (Pine Script v6). It does not generate signals from a single indicator or pattern. It requires independent confirmation across a daily bias engine, a liquidity event, and a confirmation-timeframe trigger before it will mark a Buy or Sell — and it grades every signal on how many of those layers actually aligned.
The system was built to answer one question honestly: is price sitting at a level where real institutional order flow already changed hands, or is it just sitting at a level that looks interesting on a chart? Everything in the architecture below exists to separate those two situations.
What this document is not This manual documents logic, methodology, and operating procedure. It is not a performance guarantee, a solicitation to trade, or financial advice. See Section 12. |
2. System Architecture — The Three-Layer Confluence Stack
MERIDIAN separates the trading decision into three independent layers, each running on a different timeframe and answering a different question. A signal only fires when all three agree.
Layer 1 — Daily Bias (the "should I even be looking" layer)
Anchored to 00:00 New York, this layer answers: which side of the market has the structural advantage today?
• Prior Day Value Area (POC / VAH / VAL) built from a real lower-timeframe volume profile, finalized only when the NY session rolls over — never recalculated intrabar, never repainted.
• Overnight range (00:00–07:00 NY) tracked live as the low-liquidity auction window where stops tend to rest.
• Liquidity sweep + absorption detection: did price take out the prior day's or week's high/low and then close back inside it? That is evidence of an actual transacted event, not a guess.
• Session VWAP and a cumulative volume delta (CVD) proxy confirm whether the tape is actually absorbing on the correct side.
• A composite score (-3 to +3) converts these factors into a single directional bias, filtered by an ADR-based confidence regime so thin, uninformative overnight ranges don't generate false conviction.
Layer 2 — Confirmation Timeframe (the "is this the actual moment" layer)
Two independent trigger families, each valid only when it occurs inside the live VWAP or POC confluence zone, in the direction of the active Layer-1 bias:
• Candlestick reversal detection — engulfing, pin bar / hammer / shooting star (wick-dominant, volume-spike gated), and morning/evening star. These only count as confirmation when they print at the retest, not anywhere on the chart.
• Harmonic PRZ scanning — a non-repainting, pivot-based XABCD detector for Gartley, Bat, and Crab completions, validated against Fibonacci tolerance bands and spaced to prevent the same swing structure from being counted as multiple "new" patterns.
Neither trigger is a standalone entry signal. Both exist purely to answer: has the pullback into the zone actually produced a reversal-shaped event, or is price just passing through?
Layer 3 — Signal Engine & Grading (the "how much do I trust this" layer)
A Buy or Sell only fires when Layer 1 direction, live zone-touch, and at least one Layer 2 trigger align simultaneously, subject to a bar-count cooldown that prevents the same structural event from re-firing repeatedly. Every signal is graded:
• A+ — Sweep/absorption occurred in the signal direction AND both candlestick and harmonic confirmation fired together in the same zone visit. The rarest, highest-conviction grade.
• A — Sweep/absorption occurred in the signal direction, confirmed by one Layer-2 trigger.
• B — Structural score threshold met with no confirmed sweep. Valid, but more vulnerable to trend-day failure since nothing has actually been absorbed yet.
3. Why This Qualifies as an Institutional Framework
"Institutional" is used deliberately here, not as a marketing label. It refers to a specific set of design choices that distinguish this framework from retail pattern-matching indicators:
It requires a transacted event, not a visual pattern
The sweep + absorption detection is the core of the framework. A candlestick shape alone has close to no standalone statistical edge — what gives it meaning is that it occurs at a level where a liquidity run has already been confirmed. MERIDIAN treats the candlestick and harmonic layers as confirmation of an event that already happened structurally, not as the source of the signal.
It is built around real auction-market structure
Value Area, POC, and VWAP are computed from an actual lower-timeframe volume profile and session-anchored cumulative volume — not approximated from price alone. The 00:00 NY anchor reflects how institutional desks actually delineate the trading day, rather than using the exchange's arbitrary session open.
It is non-repainting by construction, and says so where it isn't
Prior-day and prior-week reference levels use confirmed, non-repainting security calls. Where a genuine structural lag exists — harmonic pivots necessarily confirm several bars after the actual swing point, because a pivot cannot be known until price has moved away from it — that lag is disclosed rather than hidden.
It actively invalidates itself
Most retail dashboards only tell you when a bias starts. MERIDIAN tracks the overnight extremes as the structural foundation the day's bias depends on, and forcibly kills the bias the moment price closes back through the wrong side of them — sticky for the session, not a decaying score.
It grades conviction instead of pretending all signals are equal
The A+/A/B system exists because a sweep-confirmed setup and a purely structural one carry meaningfully different risk profiles. Presenting them identically would be dishonest about what the framework actually knows.
4. Best Practices for Forward Testing
Forward testing this framework correctly means testing the signal generator, not trading blind off the label. Follow this protocol:
1. Log every signal, not just the ones you take. Record: timestamp, instrument, timeframe, grade (A+/A/B), trigger source (Candle / Harmonic / Both), bias score, confidence level, entry price, stopRef, targetRef.
2. Track grade and trigger source as separate variables. Do not average them together. If B-grade signals perform statistically similarly to A-grade, the grading logic needs revisiting before you rely on it.
3. Record Maximum Adverse Excursion (MAE) before invalidation. This tells you whether stopRef is a sane risk reference or just a decorative line.
4. Test across confidence regimes separately. Do not draw conclusions from a sample dominated by one regime (e.g. mostly LOW-confidence overnight sessions) — the ADR filter exists precisely because behavior differs by regime.
5. Treat every signal as confirmation-only, never as a standalone trigger. Require agreement with your own higher-timeframe structural read (1H/4H) before execution, exactly as the framework's own zone-retest logic assumes.
6. Do not evaluate on fewer than 4–6 weeks of signals across varying volatility conditions. Two or three days of Replay-mode screenshots verifies the code isn't broken; it does not verify edge.
7. Remember the HUD panel reflects the final loaded bar only — in TradingView Bar Replay mode this means it will not update as you scrub the cursor. Use plotted historical labels, not the HUD, when auditing past signals.
On position sizing and execution MERIDIAN is a decision-support and signal-generation layer. It does not size positions, place broker-side stops, or manage exits. Every trade taken from a signal during forward testing is manually risk-managed by the operator. |
5. Optimal Timeframes
Bias Timeframe: Daily (fixed)
The Layer-1 engine is anchored to the daily/NY-midnight structure regardless of the chart timeframe you view it on. This does not change.
Confirmation Timeframe: 15-Minute to 1-Hour
This is the single most consequential timeframe decision in the system. Testing across 5M and 15M confirmed the following:
• 5-Minute: Candlestick triggers remain clean and well-spaced. Harmonic triggers become noisy — pivots confirm roughly every 25 minutes, which is thin for a structurally meaningful XABCD leg, and produced clustering even after spacing fixes were applied.
• 15-Minute (recommended default): Both trigger families produce well-isolated, non-overlapping signals with a healthy cadence — roughly one signal per major structural swing across a 1.5–2 day window in testing.
• 1-Hour: Expected to further improve harmonic selectivity at the cost of signal frequency. Not yet forward-tested; a reasonable next experiment once 15M is validated.
Entry Timeframe: 5-Minute to 15-Minute
Use a lower timeframe than your confirmation TF purely for execution precision (fill placement relative to the zone), not for generating a separate signal — the framework's signal remains the confirmation-TF event.
6. Optimal Assets & Market Conditions
Best-suited instruments
• Spot Gold (XAUUSD) and other precious metals — deep liquidity, clean overnight/London/NY session structure, reliable volume-profile behavior. This is the instrument the framework was developed and tested against.
• Major FX pairs (EURUSD, GBPUSD, USDJPY) — the NY-midnight anchor and session-based liquidity sweep logic map directly onto FX market structure.
• Major equity indices (ES, NQ) — strong respect for prior-day value area and session VWAP; overnight range behaves as designed given the defined RTH/ETH split.
Requires caution or parameter adjustment
• Low-volume altcoins or thin small-cap equities — the volume-spike gate on candlestick confirmation and the ADR-based confidence filter both depend on volume data that is meaningful and consistent. Thin or manipulated volume undermines both.
• 24/7 crypto majors (BTC, ETH) — tradeable, but the concept of an "overnight low-liquidity window" is weaker without a traditional close. Expect the confidence filter to behave differently; validate independently before relying on it.
• Any instrument during major scheduled news (NFP, CPI, FOMC) — the framework has no news-awareness layer. Sweep and volume-spike detection can misfire on news-driven spikes that are not genuine liquidity absorption.
7. Signal Grade Reference
Grade | Requires | Interpretation |
A+ | Sweep + Candle + Harmonic, same zone visit | Rarest and highest-conviction. Both confirmation families agree independently. |
A | Sweep + one Layer-2 trigger | A genuine liquidity event confirmed by one reversal signature. |
B | Structural score threshold, no sweep | Valid but unconfirmed by an actual transacted event — more vulnerable to trend-day failure. |
8. Known Limitations & Repaint Disclosures
• Harmonic pivots confirm pivotLen bars after the actual swing point by construction — this is a structural lag, not repainting, but it means the harmonic trigger is never the earliest possible signal of a reversal.
• The HUD dashboard table only updates on the chart's true final bar (barstate.islast). In TradingView Bar Replay mode it will not reflect wherever the replay cursor is positioned — use plotted labels for historical audit, not the HUD.
• No news-event awareness. Volume spikes and sweeps around scheduled high-impact news can resemble genuine absorption without being one.
• Fibonacci tolerance bands for Bat and Crab patterns overlap across a meaningful middle range of the B-leg ratio, meaning some pivot sequences can technically satisfy both. Pattern labeling in ambiguous cases should be treated as indicative, not definitive.
• This is a signal-generation and decision-support layer only. It does not size positions, place broker-side orders, or manage trade risk.
9. Input Parameter Reference
Input | Purpose |
Reference Timezone | Timezone used for the 00:00 daily anchor. Default America/New_York. |
Overnight Window Start/End | Defines the low-liquidity auction window tracked as the overnight range (default 00:00–07:00 NY). |
Lower Timeframe for Profile/CVD | Resolution of the underlying data used to build the volume profile and CVD proxy. |
Profile Bins / Value Area % | Resolution and coverage of the prior-day volume profile (POC/VAH/VAL). |
ADR Lookback / Confidence Thresholds | Defines the regime filter — how the overnight range's % of 20-day ADR maps to LOW/MEDIUM/HIGH confidence. |
VWAP / POC Zone Buffer (x ATR14) | Width of the confluence retest zones around VWAP and prior-day POC. |
Min Wick:Body Ratio / Volume Spike | Gates for candlestick confirmation — how wick-dominant and volume-confirmed a reversal candle must be. |
Zigzag Pivot Length / Fib Tolerance | Controls harmonic PRZ sensitivity — lower pivot length and tolerance increase selectivity at the cost of frequency. |
Min Bars Between Harmonic D-Pivots | Prevents overlapping XABCD windows from being counted as separate patterns. |
Minimum |Bias Score| to Arm | Threshold on the composite daily score required before a signal can arm. |
Signal Cooldown (bars) | Minimum bar spacing between any two fired signals, regardless of source. |
10. Version History
Version | Change |
1.0 | Initial release — three-layer confluence engine, candlestick and harmonic confirmation, A+/A/B grading. |
1.0.1 | Fixed a compile error: invalid "condition ? true : na" ternary in the Buy/Sell plotshape calls. |
1.0.2 | Fixed harmonic signal clustering — added consume-on-fire logic so a completed XABCD retires immediately after producing a signal instead of refiring on every zone re-touch. |
1.1 | Added minimum D-pivot spacing guard to prevent overlapping sliding-window XABCD matches from firing as separate near-duplicate signals. |
11. Risk Disclaimer
This document and the accompanying indicator are provided for research and forward-testing purposes only. Nothing in this manual constitutes financial, investment, or trading advice. Past behavior of any signal, pattern, or backtest is not indicative of future results. Trading leveraged instruments including forex, CFDs, and futures carries substantial risk of loss and is not suitable for all investors.
PSRC MERIDIAN does not execute trades, manage risk, or size positions. All trading decisions, position sizing, and risk management remain the sole responsibility of the operator. Forward-test on a demo account before committing real capital, and consult a licensed financial professional regarding your individual circumstances.
Access the Whitepaper
Comments