A Comparative Study of Artificial Neural Networks and Support Vector Machines for predicting stock prices in National Stock Exchange of India

摘要

A system for prediction of future trend of a particular index or stock increases investment opportunity in share market. Machine learning techniques are used in developing trend prediction systems. This study compares the capability of Artificial Neural Networks (ANN) and Support Vector Machines (SVM) models for predicting the closing price of a particular stock listed in National Stock Exchange (NSE) of India. Time series data of two companies, namely, State Bank of India (SBIN) and Larsen & Toubro Limited (LT), are chosen as case studies. Principal Component Analysis (PCA) was used to convert high dimensional & correlated data to low dimensional & linearly uncorrelated data. The PCA data was used with ANN to obtain better results. For SBIN, models were trained for 2012 and tested for 2013 data. For LT, models were trained for 2015 and tested for 2016 data. The computational results indicate that SVM yielded better results compared to ANN with PCA.