Machine Learning Techniques For Stock Market Prediction.Acase Study Of Omv Petrom

The research reported in the paper focuses on the stock market prediction problem, the main aim being the development of a methodology to forecast the OMV Petrom stock closing price. The methodology is based on some novel variable selection methods and an analysis of neural network and support vector machines based prediction models. Also, a hybrid approach which combines the use of the variables derived from technical and fundamental analysis of stock market indicators in order to improve prediction results of the proposed approaches is reported in this paper. Two novel variable selection methods are used to optimize the performance of prediction models. In order to identify the most informative time series to predict a stock price, both methods are essentially based on the general forecasting error minimization when a certain stock price is expressed exclusively in terms of other indicators. After the variable selection is over, the forecasting is performed in terms of the historical values of the given stock price and selected variables respectively. The performance of the proposed methodology is evaluated by a long series of tests, the results being very encouraging as compared to similar developments.