Testing Fourier and Spiking Neural Network Trading Modes

Summary

This article evaluates a trading robot that combines a discrete Fourier transform with a leaky integrate-and-fire spiking neural network. The Fourier component identifies the strongest frequency in a rolling window and maps its phase to a directional score; the neural component accumulates weighted bullish or bearish input across completed bars until a threshold is reached. Seven operating modes apply these ideas to price, MACD, RSI, price action, or combinations of them.

The author reports forward-walk tests in which parameters are optimized on the first two thirds of each test window and evaluated on the remaining third. Five settings finish positively and two lose, but the tests vary symbol, timeframe, and window at once, so they cannot isolate which mode or input caused the results. The article treats these outcomes as research leads, not proof of a durable edge, and recommends rolling tests with frozen parameters, harsher execution costs, and comparisons that remove the neural gate before considering live use.

Key ideas

• The Fourier component tracks the strongest periodic structure in a rolling sample, which may change as observations enter and leave the window.
• The spiking network accumulates directional stimulation over completed bars and signals when its threshold is reached.
• Seven modes route price and indicator inputs through different combinations of spectral analysis and neural state.
• The reported forward checks split each test window into an optimization segment and a reserved evaluation segment.
• Because symbol, timeframe, and test period change together, the mixed results do not establish a general or durable edge.

Tags

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