Forex Market Analysis Using Deep Learning Approaches

Abstract

This paper compares the effectiveness of various deep learning models which includes LSTM (Long-Short Term Memory) and GRU (Gated Recurrent Unit) models. These models use three exchange currency pairs named Euro to US Dollar, British Pound to US Dollar, and Indian Rupee to Japanese Yen for the purpose of training and performance comparison. The analysis is conducted daily according to time zones. Mean Square Error (MSE), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE) performance measures are used to compare different models. According to the observations, the GRU model outperformed the LSTM model in the majority of datasets.