Automated forex trading system using stacked machine learning and technical analysis

Forex appeals to investors who enjoy the challenge of profiting from market fluctuations. However, it also carries the risk of potential losses. To mitigate these risks, many investors seek tools for predicting price movements. This thesis suggests utilizing Stacking Machine Learning Models with Deep Q-Network to generate automated Forex trading, assisting investors in decision-making and strategy planning. Our experiments involve comparing four major currency pairs using Trading Metrics to evaluate model performance. As experiment results, our model demonstrated profitability and achieved a minimum accuracy of 65% in both Short Trades Won (%) and Long Trades Won (%).