Combining Optimized Perceptrons and Neural Networks in Forex EAs

Summary

The article describes a MetaTrader 5 workflow for combining many optimized strategy parameter sets in one expert advisor. It focuses on EURUSD hourly trading and uses perceptrons built from TEMA slope inputs, with optimized weights and a threshold parameter controlling trade frequency and signal strength. The author exports selected optimization results, embeds them in the EA, assigns trades separate comments, and limits how many result sets can operate at once.

Forward tests use one year of data after optimization on a preceding three-year window, with 20 parameter sets active. Some perceptron variants show upward performance, while other configurations, including some neural-network variants, underperform or lose deposits. The author concludes that the perceptron systems may need re-optimization at least every six months and suggests checking other pairs and timeframes. The results are limited to the stated instrument, period, and setup; the article also notes that the selected optimization window is not a reliable general criterion and that portfolio use entails more optimization work.

Key ideas

• The author combines multiple optimized parameter sets in one EA and uses order comments to track each series.
• Perceptron signals use weighted TEMA slope measurements, with a threshold that affects trade count and accuracy.
• The reported forward tests differ substantially by model variant, and some neural-network configurations perform poorly.
• The author recommends periodic re-optimization and further checks across other pairs and timeframes.

Tags

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