RSI Analog Classification with Adaptive Weights and Signal Filters

Summary

This indicator builds eight features from RSI, including its level, slope, acceleration, and volatility-related measures, then compares current conditions with a stored history of labeled outcomes. A weighted Lorentzian distance selects similar bars, whose distance-weighted votes provide directional bias. An optional Fisher discriminant estimates feature importance from the banked outcomes and updates weights over time, balancing adaptation speed against stability.

Signals require a trend flip to pass separate setup-rank and confidence thresholds, with optional trend, volatility, chop, and cooldown filters. The document explains how memory depth, neighbor count, and learning sensitivity affect analog selection and signal behavior. It describes the design and tuning choices of an indicator, but supplies no independent performance results or evidence that its historical analogs predict future returns. Its outcomes depend on selected inputs, thresholds, available history, and market regime.

Key ideas

• The indicator compares a set of RSI-derived features with stored historical examples using weighted distance.
• A distance-weighted vote among nearest analogs informs its directional bias.
• An optional Fisher-based optimizer adapts feature weights, with a tradeoff between stability and responsiveness.
• Signals are filtered using rank, confidence, trend, volatility, chop, and cooldown conditions.
• The source describes configurable logic but provides no independent evidence of predictive performance.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.