Combining Deep Learning, Heiken Ashi, and Technical Indicators for Forex

Summary

The document outlines an experimental EURUSD workflow that trains a CNN–LSTM model on scaled closing prices and exports it to ONNX for use in MetaTrader 5. It describes a historical-data split for training and evaluation, dropout and early stopping during model fitting, and RMSE as a prediction error measure. A separate strategy test combines model output with Heiken Ashi candles and indicators including PSAR, SMA, RSI, and ATR to assess trade decisions.

The article presents the approach as a rapid experiment rather than a complete trading system. It reports modest, positive backtest outcomes and describes the EA as having low drawdown, but the supplied text omits much of the strategy code and detailed test conditions. The evidence therefore cannot establish robustness, profitability after costs, or performance out of sample. Its own conclusion treats the results as preliminary and calls for further refinement; model predictions are not reliable forecasts by themselves, and risk management remains necessary.

Key ideas

• A CNN–LSTM model is trained on scaled historical EURUSD closing prices and exported in ONNX format for MetaTrader use.
• The described model uses a chronological training and testing split, dropout, early stopping, and RMSE evaluation.
• The strategy experiment combines model predictions with Heiken Ashi candles and PSAR, SMA, RSI, and ATR indicators.
• The reported backtest is characterized as modest and cautious, while incomplete code and limited test details constrain conclusions about robustness.

Tags

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