Forecasting Stock Market Movements Using Various Kernel Functions in Support Vector Machine

Abstract

In stock market forecasting achieving good prediction accuracy is always been a highly challenging task for researchers and financial analyst. Forecasting stock market needs to deal with the most volatile, non-parametric and nonlinear data sets. Also there are various factors that may affect the growth of stock market. So in order to make a good stock market forecasting system we need to use all the parameters that may affect the market volatility. Support Vectors Machine (SVM) have been found to be one of most efficient machine learning algorithm in modeling stock market prices and movements. Researchers are using these classification algorithms for so many years and have got a good predictive accuracy. Here in our research we have used SVM algorithm to making prediction for CNX NIFTY index value. In our experiment we have compared prediction accuracy for various Kernel Types of SVM.