Predicting Stock Market Movement: Trend Deterministic Data

This study focuses on the estimation of negative and positive movements of BIST 100 stock index. The predictive performances of artificial neural network, support vector machine and naive Bayes algorithm are compared. The analyzes are carried out in two stages. In the first stage, nine technical indicators to be used as input in the estimation models are calculated by using the stock index, opening, closing, highest and lowest prices. In the second stage, the trend-setting dataset is used as input and the predictions are made by using three selected machine-learning algorithms. The BIST 100 data set includes the daily closing prices covering the range of 2009-2018. With the analysis, it is concluded that the support vector machines algorithm is the best classifier. In addition, comparisons with previous similar studies and the effects of both the data set used and the prediction models are discussed