Comparative Analysis of Long-Short Term Memory, Gated Recurrent Unit, and eXtreme Gradient Boosting for Forex Prediction: A Deep Learning Approach

The Foreign Exchange (Forex) market is a dynamic arena where fortunes are won and lost in an instant inside the ever-changing financial markets, which are typified by the complex dance of global currencies. Accurate forecasting is essential for financial success in the volatile Forex market. This paper supports the use of cutting-edge machine learning methods in financial forecasting, with a particular emphasis on four currency pairs, namely, EUR/USD, GBP/USD, AUD/USD, and NZD/USD. The aim of paper is to conduct a comparative analysis of three models for Forex prediction: Long-Short Term Memory (LSTM), Gated Recurrent Unit (GRU), and eXtreme Gradient Boosting (XGBoost). These models are then thoroughly assessed using a wide range of measures, such as the Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), R2 Score, and Explained Variance Score, in addition to the Sharpe Ratio for risk-adjusted returns. The results not only confirm that LSTM is a reliable tool for financial forecasting, but it also emphasises how important it is to use gradient boosting and deep learning techniques to improve the accuracy of market trend predictions.