Predicting Boom 1000 RSI Changes for an MQL5 Trading EA

Summary

The article explores whether forecasting changes in RSI can provide a useful proxy for forecasting direction in Deriv’s Boom 1000 synthetic market. Using 100,000 one-minute observations, it compares a direct price-direction target with a target based on whether RSI rises or falls over a 20-bar horizon. The author reports that RSI direction and price direction agree about 83% of the time, while their measured correlation is weak. Two deep neural network classifiers are compared: the direct price model reaches about 53% accuracy and the RSI model about 63%.

The RSI model is exported to ONNX and connected to an MQL5 Expert Advisor that trades in the predicted direction and closes positions when the model state changes. The author reports that tuning overfit and failed to beat the default model on unseen validation data, using time-ordered five-fold cross-validation. Thus, the better reported classification accuracy does not establish profitability, and the article acknowledges that RSI direction can mislead and that the market is difficult to separate with the examined features.

Key ideas

• The study compares direct price-direction prediction with prediction of future RSI direction.
• The author reports greater classification accuracy for the RSI target than for the price target.
• RSI and price direction agree in most observations, but the reported correlation is weak.
• Time-series validation found that tuning did not outperform the default RSI model on unseen data.
• An ONNX classifier is integrated into an EA, but the reported accuracy does not demonstrate trading profitability.

Tags

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