Building a Neural Attractor Oscillator for Overbought and Oversold Signals

Summary

The article proposes a Chaos Attractor Oscillator that treats extreme price deviations as potential mean-reversion signals. Its outline combines a neural network trained on historical prices with an estimated dynamic attractor; the oscillator expresses the current price’s percentage deviation from that estimate. Positive or negative extremes are presented as possible overbought or oversold conditions. It also discusses normalizing inputs and using a tanh activation, and suggests choosing a moving average according to estimated market predictability.

The supporting material is mainly conceptual explanation and code fragments for network initialization, a forward pass, input normalization, and oscillator calculation. It offers no empirical results, validated thresholds, detailed training procedure, or evidence that the proposed attractor forecasts prices. The discussion of fractal structure and chaos motivates the design but does not establish predictive value. Any signals therefore require independent testing and careful risk controls.

Key ideas

• The proposed oscillator measures percentage deviation between price and a neural network estimated attractor.
• The article interprets large positive and negative deviations as possible overbought and oversold conditions.
• It describes normalizing historical prices and using tanh within a simple neural network.
• The article provides conceptual rationale and code fragments, but no empirical validation of predictive performance.

Tags

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