Machine Learning Key Levels

The Machine Learning Key Levels indicator plots significant price levels on your chart by detecting historical pivot points and grouping them using agglomerative clustering. This tool provides an adaptive way to visualize price zones where historical activity has concentrated, helping traders identify potential support and resistance areas based on statistical density rather than simple price touches.

Usage

The script identifies pivot highs and lows over a specified lookback period and stores them for analysis. Using hierarchical clustering, it merges these points into distinct levels.

• Identifying Support and Resistance: The horizontal lines represent the calculated key levels. Labels next to each level indicate its "strength," which can be displayed as a percentage of total pivots or a raw count.
• Adaptive Levels: By enabling the Auto No. Levels setting, the indicator uses the silhouette method to automatically determine the most mathematically sound number of clusters for the current market structure.
• Pivot Visualization: If enabled, circles appear on the chart where pivots were detected, color-coded to match the cluster (level) they belong to. This helps visualize how the machine learning model grouped specific price actions.

Details

Agglomerative clustering is a bottom-up machine learning method. It starts by treating every detected pivot as its own cluster and then iteratively merges the closest pairs based on "average linkage" (the average distance between all points in two clusters).

Unlike K-means clustering, which requires a pre-defined number of clusters (K), this hierarchical approach allows for a more flexible interpretation of price data. The script utilizes the Silhouette Method to score different cluster configurations; a higher silhouette score indicates that the levels are well-separated and that price points within each level are tightly grouped, signifying a "cleaner" and potentially more reliable key level.

Settings

Clustering

• Pivot Length: The number of bars required on either side of a high or low to confirm a pivot point.
• Pivot Memory: The maximum number of historical pivots stored in memory for the clustering engine. Higher values provide more historical depth but may impact performance.
• Number of Levels: The fixed number of key levels to display when the automatic selection is disabled.
• Auto No. Levels: When enabled, the script automatically selects the optimal number of levels using the silhouette method.

Display

• Show Pivot Circles: Toggles the visibility of markers at the original pivot locations.
• Level Glow: Adds a visual glow effect to the horizontal lines to highlight the zones.
• Display Strength As: Determines whether the level strength labels show a percentage of total pivots or the raw pivot count.

Metrics Table

• Show Metrics Table: Displays a table on the chart containing data such as the total pivot count, cluster count, and the silhouette score.
• Table Text Size / Position: Customizes the appearance and location of the metrics information.

FAQ

How do I interpret the strength of a level?

The strength is determined by the number of pivots that fall within that specific cluster. A higher percentage or count indicates that the price has reacted to that specific zone more frequently in the past.

Why do the levels move or change?

As new price data is formed and new pivots are detected, the clustering engine re-calculates the optimal averages. Because the "Pivot Memory" is a sliding window, older pivots are dropped as new ones are added, causing the levels to adapt to recent market structure.

How can I access the Machine Learning Key Levels?

You can get access on the LuxAlgo Library for charting platforms like TradingView, MetaTrader (MT4/MT5), and NinjaTrader for free.