Regime Pulse AI

Sandra Wakefield
24 hours ago
9 min read

Institutional-style regime, trend, continuation and liquidity signal framework

Primary qualified market	IC Markets XAUUSD and Metals
Primary execution timeframe	5 minutes
Default regime timeframe	60 minutes
Signal outputs	Unified BUY / SELL
Production signal locks	Markov 50–96 \| D1 bias ±0.30 \| MA separation ≤1.50 ATR

Forward-test / research use. Validate broker costs, market fit and execution before live deployment.

Download the Regime Pulse Strategy

Download the Regime Pulse Indicator with Signals

Download the Whitepaper:

System purpose and design philosophy
Core architecture and signal logic
Unified BUY and SELL signal construction
Why the framework is institutional-style
Signal dashboard and interpretation

6. Best assets, timeframes and sessions

Recommended operating practices
Risk, limitations and failure conditions
Forward-testing protocol

1. System Purpose and Design Philosophy

RegimePulse™ is a multi-layer signal framework designed to identify higher-quality continuation and liquidity-reversal opportunities only when the broader market state, completed daily bias and higher-timeframe regime are aligned. The indicator is intentionally selective: it is built to avoid treating every moving-average crossover or liquidity sweep as tradable.

The architecture combines three decision horizons: a completed daily directional bias, a completed higher-timeframe Markov-style regime state, and a lower-timeframe execution model. The indicator then qualifies entries using trend structure, continuation quality, moving-average separation, and institutional liquidity behavior.

Design principles

· Quant first: regime and directional context are evaluated before entry confirmation.

· Completed higher-timeframe data: daily and Markov inputs use prior completed bars with lookahead disabled.

· Selective admission: a signal is emitted only after multiple independent conditions agree.

· One-bar signal pulse: unified BUY/SELL alerts fire once when a new qualified condition appears.

· Fail-safe ambiguity handling: if BUY and SELL conditions appear simultaneously, both are suppressed.

· Production locks: the qualified long architecture preserves the 50–96 Markov percentile band, D1 bias floor of +0.30 and MA separation cap of 1.50 ATR.

Important qualification note

The long-side architecture is derived from the production-qualified strategy development path. The SELL side in the indicator is a mirrored bearish signal model for signal research and forward testing. It should not be treated as having the same historical qualification as the long-side executable strategy until separately validated.

2. Core Architecture and Signal Logic

2.1 Completed-D1 institutional bias

The daily bias score combines three completed-day components: normalized momentum, trend slope, and location within a rolling daily range. The components are weighted and compressed into a score between approximately -1 and +1.

Component	Purpose	Default role
Momentum	Measures directional change versus daily ATR	40%
Trend	Measures completed-D1 EMA slope versus daily ATR	35%
Range location	Measures where the prior close sits inside the recent daily range	25%

Signal lock: long setups require dailyBiasScore ≥ +0.30. Mirrored short setups require dailyBiasScore ≤ -0.30.

2.2 Higher-timeframe Markov regime

The regime engine evaluates completed higher-timeframe normalized return and EMA slope, classifying the market into bullish, neutral or bearish states. Transition counts estimate persistence from the current state, creating bull and bear edge measures. A rolling percentile rank then places current edge quality relative to recent qualified regime observations.

· Default Markov timeframe: 60 minutes.

· Bull regime requires positive normalized return and positive normalized slope above configured thresholds.

· Bear regime mirrors the same logic in the opposite direction.

· The production admission band accepts edge percentiles from 50 through 96.

· The upper cap is intentional: extremely saturated trend states are treated as potentially late-stage rather than automatically superior.

2.3 Moving-average continuation engine

The execution engine uses a fast MA, slow MA and structural trend MA. Crossovers arm setups, but a crossover alone is not a trade. The system can operate in Immediate, Pullback or Breakout mode; the default architecture favors pullback continuation.

· Trend alignment validates price and, optionally, fast/slow averages against the structural trend MA.

· Pullback mode waits for price to revisit the fast/slow MA zone.

· A microstructure break confirms renewed directional expansion.

· MA separation must remain ≤1.50 ATR to avoid late overextended entries.

· Continuation quality combines efficiency, body displacement and close location.

2.4 Institutional liquidity engine

Liquidity signals monitor Asian range extremes, London range extremes, previous-day high/low and previous-week high/low. A liquidity setup requires an actual sweep beyond a reference level, a reclaim back through that level, sufficient candle quality, and directional agreement with daily bias and Markov regime.

· Sweep depth is normalized by ATR so the definition adapts to volatility.

· Confluence counts how many institutional reference levels cluster near the sweep.

· Bullish liquidity signals seek downside liquidity capture followed by reclaim.

· Bearish liquidity signals seek upside liquidity capture followed by rejection.

3. Unified BUY and SELL Signal Construction

Unified BUY

A BUY candidate is produced when either the qualified MA-long model or qualified long-liquidity model is active.

· Completed-D1 bias score ≥ +0.30.

· Bullish Markov regime with percentile between 50 and 96.

· Trend and continuation-quality filters pass.

· MA continuation: valid crossover/setup state, pullback or breakout confirmation, and MA separation ≤1.50 ATR.

· Liquidity continuation/reversal: valid downside sweep + reclaim with required confluence and trend alignment.

· Signal must occur on a confirmed bar.

Unified SELL

A SELL candidate mirrors the framework on the bearish side.

· Completed-D1 bias score ≤ -0.30.

· Bearish Markov regime with percentile between 50 and 96.

· Bearish trend and continuation-quality filters pass.

· Mirrored MA continuation or upside-liquidity sweep/rejection qualifies.

· Signal must occur on a confirmed bar.

· SELL is currently best treated as forward-test / research until separately performance-qualified.

Alert behavior

The indicator exposes two TradingView alert conditions: RegimePulse™ BUY and RegimePulse™ SELL. Signals are one-bar pulses. If both directions occur on the same bar, the indicator suppresses both to avoid contradictory automation.

4. Why the Framework Is Institutional-Style

The term institutional-style describes the architecture, not ownership of proprietary institutional models. RegimePulse™ does not claim to replicate the private systems of any hedge fund, bank or market maker. Its institutional character comes from the way it organizes information and controls signal admission.

Institutional characteristic	How RegimePulse™ implements it
Multi-horizon decision stack	Daily bias → HTF regime → execution-timeframe trigger.
Regime-aware execution	Signals are conditioned on directional state and persistence, not only price pattern.
Liquidity references	Asian/London extremes plus prior day/week levels are treated as liquidity sources.
Volatility normalization	ATR-normalized sweep depth, displacement and MA separation make logic scale-sensitive.
Admission control	Multiple gates reduce participation in weak or saturated states.
Causal MTF handling	Completed D1/HTF observations with lookahead disabled reduce repaint risk.

The practical difference from a retail crossover indicator is that RegimePulse™ does not treat the trigger as the edge. The edge hypothesis is the conditional combination of regime, directional bias, structure, continuation quality and liquidity context.

5. Signal Dashboard and Interpretation

Field	Interpretation
MARKET	Current TradingView ticker.
EXECUTION	Shows whether the chart is on the qualified 5m execution timeframe.
D1 BIAS	Bullish / Bearish / Neutral completed-daily directional state.
BIAS SCORE	Continuous daily-bias strength.
MARKOV STATE	Current completed-HTF bull / bear / neutral state.
EDGE %ILE	Current regime-edge percentile relative to rolling qualified observations.
MA SEP	Fast/slow MA distance expressed in ATR.
CONTINUATION	Current directional continuation-quality score.
SIGNAL SOURCE	MA, LIQ, or MA + LIQ.
UNIFIED SIGNAL	BUY, SELL, or STANDBY.
CONTRACT	Timeframe compatibility status.

6. Best Assets, Timeframes and Sessions

Source-qualified configuration: the strategy development and beta workflow were centered on IC Markets XAUUSD using a 5-minute execution chart and 60-minute Markov regime. That is the only market/timeframe combination that should be described as qualified from the current system history.

Recommended deployment hierarchy

Priority	Asset class / examples	Execution TF	Status
1	Gold / XAUUSD	5m	Primary qualified use
2	Major equity indices / NAS100, US500, GER40	5m–15m	Recommended forward-test only
3	BTCUSD / ETHUSD	5m–15m	Recommended forward-test only
4	Major FX / EURUSD, GBPUSD, USDJPY	5m–15m	Recommended forward-test only
5	WTI / Brent crude	5m–15m	Recommended forward-test only

Why these markets are plausible candidates: the architecture benefits from liquid instruments with recurring session transitions, observable prior-day/week liquidity, meaningful intraday range expansion and enough volatility for ATR-normalized structure to matter. This is a model-based recommendation, not a claim of existing validation.

Best timeframe stack

· Execution: 5m is the primary qualified setting.

· Regime: 60m default Markov timeframe.

· Directional bias: completed D1.

· For slower assets or less active traders: 15m execution can be researched, but must be re-qualified because signal frequency and MA/separation behavior will change.

· Avoid 1m for production unless specifically re-engineered: microstructure noise, spread/slippage and false sweeps become materially larger.

Session practices

· Gold: prioritize London open, London/New York overlap and early New York where liquidity and displacement are typically strongest.

· FX: prioritize London and London/New York overlap.

· US indices: prioritize cash-session open and first 2–3 hours of New York.

· Crypto: use liquidity/event windows rather than assuming 24-hour uniform quality.

7. Recommended Operating Practices

Before the session

12. Confirm the correct symbol and execution timeframe.

13. Check completed-D1 bias direction and strength.

14. Check Markov state and edge percentile.

15. Mark prior day/week extremes and session liquidity.

16. Review scheduled high-impact macro events for the traded asset.

17. Confirm broker spread/slippage conditions are normal.

When a signal fires

· Treat the signal as a qualified opportunity, not an instruction to ignore execution conditions.

· Prefer signals where D1 bias, Markov state, MA structure and liquidity context all tell the same story.

· Avoid chasing if market price has moved materially beyond the signal bar before execution.

· Do not widen stops simply to preserve a trade idea.

· Do not override the 1.50 ATR separation logic by manually chasing overextended price.

Risk practice

· Use fixed fractional risk or a volatility/structure-based risk model.

· Cap daily loss, consecutive losses and trades per day.

· Use broker-side protective stops for automation.

· Forward-test real spread, slippage and commission before meaningful capital is deployed.

· Treat correlated signals across gold, indices and USD pairs as portfolio concentration, not independent risk.

8. Risk, Limitations and Failure Conditions

· No indicator eliminates market risk; a high historical PF does not guarantee future profitability.

· The long side has the strongest development evidence in the current system history; SELL is mirrored and must be independently validated.

· The Markov engine is a simplified state-transition model, not a hidden-state HMM or proprietary institutional regime engine.

· Liquidity sweeps are price-action proxies for liquidity interaction; the indicator does not use centralized futures order-book data or true dealer inventory.

· TradingView backtests can differ from broker fills. Spread, slippage, commissions and execution delay must be measured.

· News shocks can invalidate normal ATR, regime and reclaim behavior.

· Changing symbols, timeframes, MA settings or regime parameters invalidates prior qualification unless re-tested.

Failure conditions that should stop trading

· Persistent cost-adjusted PF below 1.20 over a meaningful forward sample.

· Live/OOS PF below 1.10.

· Max strategy drawdown above 5%.

· Material broker/local position mismatch.

· Any unprotected live position.

· Sustained execution costs materially above the qualification envelope.

· Repeated signal clustering caused by abnormal volatility or data-feed anomalies.

9. Forward-Testing Protocol

Beta testing should validate execution quality and regime behavior rather than searching for a new best parameter set.

Metric	What to record	Why it matters
Signal timestamp	Exact bar/time of BUY or SELL	Verifies alerts and reproducibility
Signal source	MA / LIQ / MA+LIQ	Identifies strongest setup family
D1 bias / Markov	State and score at signal	Validates regime conditioning
MA separation	ATR-normalized value	Monitors overextension
Spread / slippage	Observed broker values	Determines live cost drag
MFE / MAE	Maximum favorable/adverse excursion	Evaluates entry quality
Outcome in R	Net result after costs	Creates comparable expectancy data

Minimum beta objective: accumulate enough independent signals to evaluate cost-adjusted expectancy separately for BUY and SELL. Do not pool mirrored SELL results with historically qualified BUY results when deciding production readiness.

10. Quick-Reference Checklist

A+ BUY profile

· D1 bias ≥ +0.30 and clearly bullish.

· Bull Markov state with edge percentile 50–96.

· Price structurally aligned above trend MA.

· MA separation ≤1.50 ATR.

· Continuation quality strong.

· Pullback + micro-break or institutional downside sweep + reclaim.

· Reasonable spread/slippage and no immediate high-impact event risk.

A+ SELL profile

· D1 bias ≤ -0.30 and clearly bearish.

· Bear Markov state with edge percentile 50–96.

· Price structurally aligned below trend MA.

· MA separation ≤1.50 ATR.

· Bearish continuation quality strong.

· Pullback + micro-break or institutional upside sweep + rejection.

· Forward-test status respected until the short side is independently qualified.

Do not trade when

· Bias and Markov regime conflict.

· Regime edge is outside the production band.

· MA separation is overextended.

· The chart timeframe is incompatible with the selected regime timeframe.

· The signal fires into abnormal spread, illiquidity or major unscheduled volatility.

· You are increasing risk to compensate for a recent losing streak.

11. TradingView Indicator Description

RegimePulse™ Unified Signals is a non-repainting, multi-timeframe market-regime and liquidity indicator built for selective intraday execution. It combines completed-D1 institutional bias, a completed-HTF Markov regime engine, ATR-normalized trend/continuation filters, and institutional liquidity sweeps from Asian/London ranges plus previous day/week extremes. The indicator outputs unified BUY and SELL signals only after regime, bias, structure and confirmation align. Production long-side locks retain the validated 50–96 Markov percentile band, D1 bias ≥0.30 and MA separation ≤1.50 ATR. Includes unified alerts, a high-tech signal HUD, MA structure, and liquidity levels. Primary qualified use: XAUUSD 5m with 60m regime context. SELL signals are mirrored for forward testing and should be independently

validated.