AetherEdge Multi-Feature Neural Divergence Hunter

The AetherEdge Multi-Feature Neural Divergence Hunter indicator provides an advanced divergence detection engine that monitors six independent oscillators and utilizes neural weight learning to prioritize the most accurate signals. This tool tracks regular and hidden divergences simultaneously, applying dynamic weighting and confluence filters to identify high-probability reversal and trend-continuation points.

Usage

The Usage section describes how the script can be used to identify market transitions and trend continuations. Users can monitor signals across multiple timeframes and assets, with the neural weighting system automatically adjusting to specific market conditions.

Divergence Identification

The script detects two primary categories of divergence across six oscillators (RSI, MACD, Stochastic, CCI, MFI, and Momentum):

• Regular Divergence: Used to identify potential trend reversals.
• Hidden Divergence: Used to identify potential trend continuations during pullbacks.

Confluence and Visual Encoding

The indicator uses a 3D visual encoding system to communicate signal strength and type:

• Color: Identifies the direction and type (Regular, Hidden, or Mixed).
• Size: Represents the confluence count; larger icons indicate that more oscillators have confirmed the divergence.
• Shape: Different shapes distinguish between regular (triangles), hidden (arrows), and mixed signals (labels).

Neural Oscillator

The composite Neural Oscillator (nnOsc) displays a weighted average of all active features. Levels above 35 suggest overheated conditions, while levels below -35 suggest bottoming zones. Crosses through the zero line can be used as directional confirmation for detected divergences.

Details

The script operates as a parallel processing system for six normalized oscillators.

• Feature Normalization: Oscillators like MACD and Momentum are normalized to a 0–100 scale over a 100-bar range to ensure they are comparable with bounded oscillators like RSI and Stochastic.
• Neural Weight Learning: The system evaluates every signal after a user-defined "Reward Horizon." If the price moves in the predicted direction, the weight for that specific oscillator increases; if it moves against the signal, a penalty is applied.
• Softmax Normalization: The raw weights are processed through a Softmax function. The "Temperature" setting determines if the script focuses heavily on the single best-performing oscillator or maintains a balanced view of all features.
• Pivot Management: The script utilizes a pivot-based detection method with a fixed lookback, ensuring consistency and eliminating dynamic history access issues.

Settings

• Pivot Lookback: Sets the number of bars required to the left and right to confirm a price pivot.
• Min/Max Bars Between Pivots: Defines the window of time in which a divergence must form to be valid.
• Feature Selection: Toggle switches for RSI, MACD, Stochastic, CCI, MFI, and Momentum.
• Learning Rate: Controls how aggressively the neural weights are updated based on successful or failed signals.
• Reward Horizon (bars): The duration the script waits to determine if a divergence signal was accurate.
• Softmax Temperature: Adjusts weight distribution; lower values concentrate weight on top-performing features, while higher values distribute weight more evenly.
• Min Confluence to Signal: The minimum number of oscillators that must show divergence simultaneously to trigger a visual alert.

FAQ

How do I interpret a "Mixed" signal? A Mixed signal occurs when both Regular and Hidden divergences are detected simultaneously across different oscillators. This often indicates a transitional or unstable market phase where caution is advised.

Why do signals appear several bars after a price peak or trough? The indicator relies on pivot points, which require a "Pivot Lookback" (default 5 bars) to confirm. Therefore, signals are plotted once the lookback period has passed and the pivot is confirmed.

How can I access the AetherEdge Multi-Feature Neural Divergence Hunter? You can get access on the LuxAlgo Library for charting platforms like TradingView, MetaTrader (MT4/MT5), and NinjaTrader for free.