Prediction of Istanbul Securities Exchange composite index

Abstract

This study presents a software developed by using Nested Generalized Exemplars, for predicting Istanbul Securities Exchange Composite Index. Information reflected in the past values of frequently used monetary variables are used to predict stock returns. Daily returns of the composite index are predicted by using: Central Bank effective selling price of US Dollar and Deutsche Mark, Istanbul Tahtakale closing selling price of Turkish Republic gold coin and one ounce of gold, Commercial Banks (İş Bank, Akbank, Yapı Kredi Bank, and Ziraat Bank) 3-month average deposit rate and 3-month Government bond interest rates. Data prior to the dates on which the predictions are made are used to learn the forecasting power of variables on composite index and to generate the appropriate rules. The results reveal that the information reflected in the past prices of the variables have significant effects on the ISE composite index.