AetherEdge AI Dynamic Trendline Architect

The AetherEdge AI Dynamic Trendline Architect indicator provides an automated trendline system powered by a multi-layer AI architecture that evaluates line quality through machine learning and structural validation. It aims to filter market noise by identifying and maintaining only high-probability support and resistance lines based on historical patterns and real-time performance.

Usage

The Usage section describes how the script can be used to identify key market structures and potential breakout points.

• Trend Reversal Identification: High-score lines (typically ≥ 0.70) indicate robust support or resistance levels. Users can look for long entries at support lines and short entries at resistance lines.
• Breakout Signals: When price breaches an active line beyond the user-defined ATR buffer, the line changes color (based on Bullish Break Color or Bearish Break Color), signaling a potential trend-following entry.
• Support/Resistance (S/R) Flips: Users can monitor lines that were recently broken to see if they transition from resistance to support (or vice versa), providing high-conviction entry zones.
• Visual Lifecycle: Lines fade in intensity as they age if Fade by Age is enabled, helping users prioritize fresher structural data. Strength labels provide a real-time confidence score for each line.

Details

The indicator utilizes a four-layer decision system to manage trendlines:

1. Pivot & Feature Engine: Detects swing highs/lows using ta.pivothigh/low and extracts five dimensions of data: ATR-normalized slope, duration, volatility ratio, RSI state, and angle.
2. KNN Pattern Memory: Matches current candidate lines against a database of historical patterns. It uses Euclidean distance to find the most similar historical outcomes and produces a consensus score.
3. Neural Quality Network: An online neural network predicts line quality. It updates its weights in real-time based on actual market outcomes (bounces or breaks) to refine its accuracy over time.
4. Structural Prior Validation: A physical validation layer that counts actual price touches and violations. It calculates a score based on how many times the price respected the line versus how many times it was violated.

The final output is a hybrid fusion of the Neural Network prediction and the Structural Prior. Only lines exceeding the Min Quality Score are rendered on the chart.

Settings

Pivot Engine

• Pivot Length: The number of bars required to confirm a swing high or low.
• Max Active Trendlines: Limits the maximum number of lines drawn on the chart simultaneously.
• Scan-back Pivots: Determines how many previous pivots the engine scans to attempt pairing with a new pivot.

KNN Pattern Memory

• K Neighbors: The number of historical neighbors used for the KNN consensus score.
• Pattern Memory Size: The maximum number of historical patterns stored in memory.

Neural Quality Network

• Learning Rate: Controls how aggressively the neural network updates its weights based on new outcomes.
• Epochs/event: The number of training iterations performed whenever a line lifecycle ends.

Line Lifecycle

• Line TTL (bars): The "Time To Live" for a line before it expires automatically.
• Min Quality Score: The threshold a line must reach to be displayed on the chart.
• Break Buffer (× ATR): The distance price must move beyond a line (normalized by ATR) to trigger a "break" signal.
• Structural Prior Mix: Blends the AI prediction with physical structure (0 = 100% AI, 1 = 100% physical touches).

FAQ

How do I adjust the indicator if there are too many lines on my chart?
Increase the Min Quality Score (e.g., to 0.70) or reduce the Max Active Trendlines setting to ensure only the highest-conviction levels are shown.

Why do lines occasionally disappear or reset?
Trendlines have a Line TTL (bars) setting; once a line reaches this age, it is removed. Additionally, the KNN memory and NN weights reset if the indicator is reloaded or settings are changed.

How can I access the AetherEdge AI Dynamic Trendline Architect?
You can get access on the LuxAlgo Library for charting platforms like TradingView, MetaTrader (MT4/MT5), and NinjaTrader for free.