Stock market index predicti on using machine learning techniques. A case study of bucharest stock exchange
Financial markets are complex systems and their behavior is characterized as non-stationary and volatile. In the presented paper, the potential of the machine learning techniques, in particular, adaptive network-based fuzzy inference system (ANFIS) model and Support Vector Machines (SVM) are investigated in forecasting task of stock market data. The performance of proposed models are evaluated with test data of Bucharest Stock Exchange (BVB) and results obtained are compared in terms of well known evaluation metrics. The comparative analysis leads to the conclusion that both models can be successfully applied to forecast the high-frequency stock data while SVM outperforms the ANFIS model in prediction accuracy.
Index Terms - ANFIS, Fuzzy neural network, Support Vector Machines, Time series prediction, stock market