Forcasting prices of futures contracts using artificial neural networks: An application on TURKDEX

The prediction of financial time series is an important concept. The purpose of this study is to perform an application of Artificial Neural Networks (ANN) on financial sector. The application to be carried out, I used artificial neural networks architecture to model and predict financial time series. The performance of ANN models have been measured by the results of Mean Squared Error (MSE) and Regression. Data have been collected from the official web site of Turkish Derivatives Exchange (TURKDEX). Index Futures Contracts and Currency Futures Contracts which are in TURKDEX have been used in this study. The contracts to be used in the application contain TurkDEX-ISE 30 Index Futures Contract, TurkDEX-ISE 100 Index Futures Contract, TurkDEX-TRYUSDollar Futures Contract, and TurkDEX-TRYEuro Futures Contract. 1486 days of data between 4 February 2005 and 31 December 2010 used in this study.In the current study, the basic neural model was a feed-forward architecture using the back propagation algorithm. It also performed to be minimum number of errors caused by back propagation algorithm. As a result of analyses of the established model for prediction of the contracts, it shows that ANNs method which is a nonlinear modeling technique has an effective performance.