Prediction Models for Forex Data Exchange System

Foreign exchange prediction is of important interest to investors and individual traders in financial industries in other to maximize profits and reduces  losses. However owing to some factors and the non- linearity of the FX markets especially in a developing  economy like Nigeria, generating suitable, accurate and appropriate FX predictions becomes difficult for the traders of the market. This study utilized models that include various machine learning algorithm over a trend analysis and pattern of its prediction. The model results on the currency pair of United States(USD) over Nigeria Naira (NGN) using Root Mean Squared Error (RMSE), Mean Absolute Error(MAE), Mean Square Error (MSE), and R-square (R2) showed GRU performed better in predicting the trend and we therefore considered it best fit for the forecast. The result showed high prediction over ANN and LSTM, with RMSE, MAE, MSE, and R2 values of 0.112, 0.075, 0.013, and 0.969. Keywords:- Forex, ANN, LSTM, GRU MAE, MSE.