Machine Learning RSI ║ BullVisionOverview:
Introducing the Machine Learning RSI with KNN Adaptation – a cutting-edge momentum indicator that blends the classic Relative Strength Index (RSI) with machine learning principles. By leveraging K-Nearest Neighbors (KNN), this indicator aims at identifying historical patterns that resemble current market behavior and uses this context to refine RSI readings with enhanced sensitivity and responsiveness.
Unlike traditional RSI models, which treat every market environment the same, this version adapts in real-time based on how similar past conditions evolved, offering an analytical edge without relying on predictive assumptions.
Key Features:
🔁 KNN-Based RSI Refinement
This indicator uses a machine learning algorithm (K-Nearest Neighbors) to compare current RSI and price action characteristics to similar historical conditions. The resulting RSI is weighted accordingly, producing a dynamically adjusted value that reflects historical context.
📈 Multi-Feature Similarity Analysis
Pattern similarity is calculated using up to five customizable features:
RSI level
RSI momentum
Volatility
Linear regression slope
Price momentum
Users can adjust how many features are used to tailor the behavior of the KNN logic.
🧠 Machine Learning Weight Control
The influence of the machine learning model on the final RSI output can be fine-tuned using a simple slider. This lets you blend traditional RSI and machine learning-enhanced RSI to suit your preferred level of adaptation.
🎛️ Adaptive Filtering
Additional smoothing options (Kalman Filter, ALMA, Double EMA) can be applied to the RSI, offering better visual clarity and helping to reduce noise in high-frequency environments.
🎨 Visual & Accessibility Settings
Custom color palettes, including support for color vision deficiencies, ensure that trend coloring remains readable for all users. A built-in neon mode adds high-contrast visuals to improve RSI visibility across dark or light themes.
How It Works:
Similarity Matching with KNN:
At each candle, the current RSI and optional market characteristics are compared to historical bars using a KNN search. The algorithm selects the closest matches and averages their RSI values, weighted by similarity. The more similar the pattern, the greater its influence.
Feature-Based Weighting:
Similarity is determined using normalized values of the selected features, which gives a more refined result than RSI alone. You can choose to use only 1 (RSI) or up to all 5 features for deeper analysis.
Filtering & Blending:
After the machine learning-enhanced RSI is calculated, it can be optionally smoothed using advanced filters to suppress short-term noise or sharp spikes. This makes it easier to evaluate RSI signals in different volatility regimes.
Parameters Explained:
📊 RSI Settings:
Set the base RSI length and select your preferred smoothing method from 10+ moving average types (e.g., EMA, ALMA, TEMA).
🧠 Machine Learning Controls:
Enable or disable the KNN engine
Select how many nearest neighbors to compare (K)
Choose the number of features used in similarity detection
Control how much the machine learning engine affects the RSI calculation
🔍 Filtering Options:
Enable one of several advanced smoothing techniques (Kalman Filter, ALMA, Double EMA) to adjust the indicator’s reactivity and stability.
📏 Threshold Levels:
Define static overbought/oversold boundaries or reference dynamically adjusted thresholds based on historical context identified by the KNN algorithm.
🎨 Visual Enhancements:
Select between trend-following or impulse coloring styles. Customize color palettes to accommodate different types of color blindness. Enable neon-style effects for visual clarity.
Use Cases:
Swing & Trend Traders
Can use the indicator to explore how current RSI readings compare to similar market phases, helping to assess trend strength or potential turning points.
Intraday Traders
Benefit from adjustable filters and fast-reacting smoothing to reduce noise in shorter timeframes while retaining contextual relevance.
Discretionary Analysts
Use the adaptive OB/OS thresholds and visual cues to supplement broader confluence zones or market structure analysis.
Customization Tips:
Higher Volatility Periods: Use more neighbors and enable filtering to reduce noise.
Lower Volatility Markets: Use fewer features and disable filtering for quicker RSI adaptation.
Deeper Contextual Analysis: Increase KNN lookback and raise the feature count to refine pattern recognition.
Accessibility Needs: Switch to Deuteranopia or Monochrome mode for clearer visuals in specific color vision conditions.
Final Thoughts:
The Machine Learning RSI combines familiar momentum logic with statistical context derived from historical similarity analysis. It does not attempt to predict price action but rather contextualizes RSI behavior with added nuance. This makes it a valuable tool for those looking to elevate traditional RSI workflows with adaptive, research-driven enhancements.
