Institutional Order Flow Signals

The Institutional Order Flow Signals indicator utilizes a Gaussian Naive Bayes classifier to analyze cumulative volume delta (CVD) divergence and momentum, providing real-time probabilistic insights into institutional market regimes.

Usage

The Usage section describes how the script can be used to identify market transitions and high-conviction entries based on volume flow analysis.

Signal Interpretation

The indicator identifies three primary market states:

• Bullish Re-alignment: Occurs when price and CVD both trend upward, signaling institutional buying pressure confirming price action.
• Bearish Re-alignment: Occurs when price and CVD both trend downward, signaling institutional selling pressure confirming price action.
• Order Flow Divergence: Occurs when price and CVD move in opposite directions, often indicating accumulation or distribution phases.

Trade Execution

• Long Signals: Triggered when the bullish posterior probability exceeds the user-defined threshold, CVD momentum is positive, and price remains above the trend EMA.
• Short Signals: Triggered when the bearish posterior probability exceeds the threshold, CVD momentum is negative, and price remains below the trend EMA.
• Visual Confidence: Signal shapes vary based on conviction levels. Circles represent high-conviction signals (Posterior > 85%), while triangles represent standard signals.

Details

The script implements a Gaussian Naive Bayes classifier natively within Pine Script® v6. It calculates three Z-score normalized features derived from a synthetic Cumulative Volume Delta (CVD) model:

1. CVD Momentum (F1): The rate of change of order flow relative to its recent baseline.
2. Price/CVD Divergence (F2): Measures the spread between price velocity and volume velocity.
3. CVD Slope (F3): A linear regression slope providing a direction-of-flow signal.

The classifier uses Welford’s Online Learning algorithm to update mean and variance statistics on every bar. This allows the model to adapt to changing market microstructure without requiring fixed lookback arrays or external data feeds. A warmup period is required to accumulate sufficient training observations before signals activate.

Settings

Naive Bayes Classifier

• Classifier Lookback: The minimum number of training bars required before signals become active.
• Entry Threshold: The minimum posterior probability required to trigger a signal (e.g., 0.70).
• Prior Bull Probability: The initial assumption of market direction before data processing (0.50 is neutral).

CVD Settings

• CVD Momentum Period: The lookback window used for calculating the CVD rate of change.
• CVD Slope Period: The regression window for determining the linear slope of the CVD.
• Z-Score Window: The period used to normalize features into standard deviations.

Signal Levels & Visuals

• SL ATR Multiplier: Sets the stop-loss distance based on market volatility.
• Risk : Reward: Determines the target take-profit line relative to the stop-loss.
• Trend EMA Period: A macro filter that restricts long signals to uptrends and short signals to downtrends.
• Posterior Band: Toggles the dynamic fill band anchored to the trend EMA, which changes opacity based on classifier confidence.

FAQ

How does the indicator calculate volume delta without bid/ask data? The tool uses a synthetic CVD formula based on the closing position within the high-low range of a bar to estimate net buying and selling pressure.

Why are there no signals at the beginning of the chart? The classifier requires a "warmup" period (defined by the Classifier Lookback setting) to train its statistical model on historical data before it can generate reliable probabilities.

How can I access the Institutional Order Flow Signals? You can get access on the LuxAlgo Library for charting platforms like TradingView, MetaTrader (MT4/MT5), and NinjaTrader for free.