Neural Network Based Forecasting of Foreign Currency Exchange Rates

The foreign currency exchange market is the highest and most liquid of the financial markets, with an estimated $1 trillion traded every day. Foreign exchange rates are the most important economic indices in the international financial markets. The prediction of them poses many theoretical and experimental challenges. This paper reports empirical proof that a neural network model is applicable to the prediction of foreign exchange rates. The exchange rates between Indian Rupee and four other major currencies, Pound Sterling, US Dollar, Euro and Japanese Yen are forecast by the trained neural networks. The neural network was trained by three different learning algorithms using historical data to find the suitable algorithm for prediction. The forecasting performance of the proposed system is evaluated using three statistical metrics and compared. The results presented here demonstrate that significantly close prediction can be made without extensive knowledge of market data. The currency exchange rates play a significant role in compulsive the dynamics of the currency market. Proper prediction of currency exchange rate is a crucial factor for the success of many business and investment firm. Although the market is well known for its unpredictability, fickleness and volatility, there are number of groups like Banks, Agency and other for predicting exchange rates using numerous techniques. There are many types of theoretical models including both time series and econometric approaches have been widely used to model and forecast exchange rates such as Autoregressive Conditional Heteroskedasticity (ARCH), General Autoregressive Conditional Heteroskedasticity (GARCH) and chaotic dynamics applied to financial forecasting. While these models may be better for a particular situation and they perform poorly in other applications. Artificial Neural Networks (ANNs) have received more attention as decision-making tools. The Artificial Neural Networks are the well-known function approximates in prediction and system modeling, has recently shown its great applicability in time series analysis and forecasting. Artificial Neural Network is more effective in describing the dynamics of non-stationary time series due to its unique non- parametric, noise-tolerant and adaptive properties. ANNs are universal function approximates that can map any nonlinear function without a prior assumptions about the data. The main purpose of this paper is to investigate the use of Artificial Neural network based methods for prediction of foreign exchange rates with accurate. We apply Back Propagation Neural Network (BPNN) for predicting currency exchange rates of Indian Rupee (INR) against four other currencies such as Pound Sterling (PS), US Dollar (USD), EURO, and Japanese Yen (JYEN). BPNN is trained with three different existing learning algorithms. Total of 80% historical exchange rates data for each of four currency rates, were collected and used as an inputs to build the prediction system in our study and then additional 20% exchange rates data were used to evaluate the model. The prediction results of all these models were compared based on three S. Kumar Chandar et.al / International Journal on Computer Science and Engineering (IJCSE)