Predicting Stock Price Movement with Classification Trees in R

Abstract

Stock price movement prediction is an interesting but highly complicated task. Many factors can cause the price to rise or fall. It turns out, that because of the high volatility of the stock market, the data mining techniques perform better than the traditional forecast methods to get a trading rule. This thesis attempts to develop four models, varying in the length of the time window, based on the CART algorithm for classification trees. Various technical indicators were chosen as features used in the proposed models. Experimental results show that models for longer time periods report high expected accuracy end perform better than models for short time periods.