A Model for Forex Market Price Prediction using Deep Learning

The forex markets are full of uncertainties that are largely influenced by the forces of demand and supply. The rates usually adjust depending on the prevailing status of the economy, politics and the monetary policies. The forex market consists of multiple dealers and online forex trading platforms. Predicting forex market prices is a complicated process and subjective in nature for forex dealers, economists and business people. The potential to make losses due to poor speculative guesses is quite high for multinational corporations located in more than one economy. The aim of this study is to develop a model for forex market price prediction. We develop the model using the Data-Driven modelling technique. We source the Central Bank of Kenya’s (CBK) historical dataset to achieve this. The dataset is divided into training and testing data by a splitting of 80-20. We construct the model by combining time-series regression with resilient backpropagation neural networks in the select currencies. The unique behavior of each of the currencies required a separate implementation of the model output. This results in increased accuracy and lower error levels, hence efficiency and optimality. Successful predictions are conducted up-to eight months forward. We realize accuracy levels ranging from 88-98% and SSE of 0.496-2.667. The wider the historical data range, the higher the accuracy and consequently, the longer the prediction horizon.