Stock Market Analysis and Prediction Using Artificial Neural Network Toolbox

Abstract

Stock exchange is an open market for the exchanging of organization chart. Shares will be considered as minor entities of a company. Such small entities can be bought by different stake holders. In general, the company holds majority of the shares. Based on different factors the price of such shares is fixed. Stock brokers may be used by some companies to buy/sell the shares. The stock holders generally give advice to companies based on the public opinion. This entails the fact that a lot of factors cause variation in prices of the stock, making assumed prediction based on insignificant set of factors often inaccurate. In this paper, we explore an approach using machine learning techniques (in particular neural networks). Stock market closing prices of varied stocks are predicted using different algorithms on the ANN toolbox in MATLAB and correlation of predicted and acquired value analysis is performed pertaining to a NARX network which is nonlinear and autoregressive, amongst other accuracy metrics and performance criteria to draw out the precision and reliability of the respective training function.