Abstract:

Forecasting of the stock exchange index future values with time series analysis has a great interest in finance field. There are different methods of time series used for the stock market index forecasting. One of these methods is Artificial Neural Networks (ANN) used in many studies in recent years. Artificial neural networks do not require prior conditions and they have a flexible modeling structure. For this reason, artificial neural networks are superior when compared with other methods of time-series. In this study, we aimed to forecast the Morgan Stanley Capital International (MSCI) Turkey indexby artificial neural network. Data period is December 1987-August 2008. We used feed-forward network modeling consisting of 12 input, 11 hidden and 1 output neurons. As performance criteria the root mean square error was selected and its value was calculated as 0.1131. As a result successful predictions were obtained

Keywords: