Goertzel Spectral Features with an Online Neural Ensemble
Summary
GoertzelBrain combines Goertzel frequency analysis with an ensemble of online-trained neural networks to turn detected cycle structure into a directional confirmation signal. It extracts cycle period, amplitude, spectral confidence and their changes, alongside price slope and volatility, then feeds these features to ten small multilayer perceptrons. Their average output is interpreted as bullish when positive and rising, or bearish when negative and falling. The indicator is presented as a filter to confirm trades generated by another system.
The article explains the architecture, MQL5 implementation, configurable spectral range and retraining behavior, and access to the confirmation value for an Expert Advisor. It illustrates the signal on a currency chart, but provides no systematic performance study or evidence of predictive edge. The author cautions that cycles can shift or disappear, neural networks may adapt to noise, and computational cost can rise in large backtests. Signals can also vary between indicator instances because their initial weights are random.
Key ideas
- Goertzel analysis supplies spectral features while neural networks learn how to interpret them in context.
- The feature set combines cycle measures with price slope and volatility.
- Ten independently initialized networks contribute to an averaged ensemble signal.
- A positive rising output confirms long direction, while a negative falling output confirms short direction.
- The indicator is a confirmation filter, and the article does not establish a reliable trading edge.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.