Application of Logistic Regression in Assessing Stock Performances

Stock market prediction pertains to predicting the future value of stock of a company or other financial instrument traded on an exchange. Though the successful prediction of a stock's future price is a complex phenomenon, even a reasonable prediction would yield significant profit and this study attempts to address this unpredictability with the help of a data mining technique. In this study, the companies listed on BSE SENSEX are chosen as representative set of companies that are most actively traded. Logistic Regression is used on various important financial ratios of these companies and certain macro financial variables to analyze which ratios are important and how they are affecting the stock prices. The proposed model results in better classification accuracy when compared to a similar study in the literature.

I. Introduction

To make a good investing decision or while exploring the investing options in stock market, it is important for the shareholders and potential investors to analyze the relevant financial information available with them. But at the same time, foreseeing the future stock performance is very difficult and complicated and as such there is no tool available in the market to predict future stock performance. But to an extent, stock performance can be analyzed on the basis of financial indicators that we can get from the annual report of the company. Company's annual report consists of huge amount of data that can be converted into key financial ratios that can help investors to do the analysis. Different users such as management bankers, creditors, shareholders and researchers use these financial ratios to project future stock price trends. The study of financial ratios emerged as a new discipline after stock market crashes in the 1990s and early 2000s in the United States and parts of Europe and southern Asia. The level of importance given to financial ratios differs from one country to another. Thus, selecting appropriate ratios is very crucial in increasing the prediction success rate.