Adaptive Learning Rates for Neural Network Trading in MQL5

Summary

This article outlines an MQL5 expert advisor that uses a neural network to generate buy or sell signals from moving averages, RSI, and ATR. The network has input, hidden, and output layers, uses sigmoid activation, and is trained through backpropagation. The proposed adaptive learning rate responds to training performance: it can increase when predictions improve and decrease when errors rise. The article also describes adjusting the hidden layer size according to market volatility and includes configurable trade size, stop loss, and take profit settings.

The implementation is presented as a framework, with code structure and functions for training and adjusting the learning rate. Although a testing section claims backtesting was performed, the supplied text gives no report figures or detailed evaluation results. It therefore does not establish that the method predicts market movements reliably or performs profitably. The author recommends further refinement, parameter tuning, backtesting, and risk controls before considering live use.

Key ideas

• The network uses moving averages, RSI, and ATR as inputs for two-class trading signals.
• Backpropagation updates weights, while an adaptive learning rate is intended to respond to training errors and accuracy.
• The proposed design changes hidden-layer size in response to market volatility.
• The article provides no specific backtest statistics in the supplied text, so trading performance cannot be assessed.

Tags

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