Stock market movement direction prediction using tree algorithms

One of the highly challenging businesses today is the task of forecasting the market movements by examining the financial time series data as correctly as possible in order to hedge against the almost incalculable risk involved and to yield better profits for investors. If there was a highly credible estimation technique available giving better results than the traditional statistical tools for financial markets, it would be a great asset for trading decision makers of all kinds such as speculators, arbitrageurs, portfolio fund managers and even individual investors. In this study CART, C4.5 and Random Forest algorithms were used to predict the movement direction of a 10 year Istanbul Stock Exchange index (XU-100). Ten technical market indicators such as momentum, MACD and RSI were used in this study as the feature set. Keywords: Price movement direction, CART, C4.5, Random Forest, forecasting, stock market