Preparing Data and Training Direction-Specific Trading Networks in Python

Summary

This article describes a neural-network trading workflow using Python, TensorFlow, and Keras, with MQL5 scripts supplying historical prices and indicator data. The system separates buy and sell examples and uses two networks for each direction: one learns to reproduce indicator-like features, while another produces a signal used by the trading strategy. The examples focus on EURUSD hourly data and describe preparing files of price-derived and technical-indicator inputs.

The resulting signals are used in an Expert Advisor whose entry levels, trading hours, and stop and target settings can be optimized. The article discusses optimizing indicator thresholds first, then timing and risk parameters, and testing the chosen settings on later data. It provides code-oriented examples and an optimization workflow, but the supplied text gives no clear quantitative performance evidence. Results from parameter optimization may not generalize, so the described system should be treated as an example requiring careful out-of-sample validation.

Key ideas

• The workflow uses MQL5 scripts to prepare price and indicator data for Python neural-network training.
• Separate buy and sell datasets feed direction-specific networks, with one network modeling indicator-like inputs and another producing strategy signals.
• The example trading system uses EURUSD hourly data and technical indicators.
• Indicator thresholds, trading hours, and stop and target levels are proposed as optimization parameters.
• The article provides an implementation example but does not establish robust out-of-sample performance.

Tags

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