Analysis of Feedforward and Recurrent Neural Network in Forecasting Foreign Exchange Rate

Abstract

: The forecasting of noisy, volatile and unstable natured foreign exchange rate has been challenging for all the researchers and business practitioners. The Artificial Neural Network has proved to be a very powerful tool for performing predictions and forecasts making it applicable for forex forecasting. In this paper we attempt to compare the performances of feed forward neural network and recurrent neural network used for forecasting the daily exchange rate of Indian Rupee against four base currencies namely, British Pound, US Dollar, Euro and Japanese Yen. This paper presents a detailed analysis over the network parameters, their role in training and their effect on the results. Overall both the networks gave good results for all the currencies with the recurrent network giving better results than feedforward.