Machine Learning based comparison of financial forecasting methods

Abstract

 Financial market forecasting is one of the happening fields and stock forecasting is considered by many as a relatively straightforward when compared to other derivatives in the market. Stocks in financial market are generally forecasted using many models from Statistics, Mathematics, Economics or machine learning in general. Three of the prevalently employed methods from machine learning domain are, Monte Carlo simulation, linear regression and Autoregressive modeling. This paper is trying to compare the efficiency of the three models by forecasting the stock prices of State bank of India using all three. To structure the whole process we employ a neural network model initially to finalize the explanatory and regressed variable for the prediction process. Then all three methods are employed for forecasting and their results are compared to arrive at a better model among the three.