Stock Predictability Using Sparse Learning Approach

Abstract

This research paper aims to predict the stock returns of the theS&P 500 companies by using Sparse Learning Approach with the help of historical stock data. In the field of finance and microeconomics, variables keep expanding every day due to different factors or strategies introduced by humans intending to make the result profitable thereby making the process more complex. In the world of complexity, sparse learning is one of a kind approach to deal with huge variable sets. Hence, the intention of the paper is to apply the sparse methodology to the financial data of S&P 500 companies and observe the applicability of the model to these types of data sets. Variable selection is an important and quite challenging task in econometric modeling. There are different types of algorithms readily available to optimize and regularize the data sets of various processes in different fields such as medicine, finance, and telecommunication sectors. However, a major problem faced by an individual while working on research is the selection of variables in the data sets to carry out the analysis. To overcome this difficulty, we are interested in utilizing sparse learning model, especially lasso regression. In this paper, we have introduced the methodology and objective behind the model by analyzing it with S&P 500 data