Prediction analysis for US stock market

Abstract

The goal of this paper is to use prediction analysis for managing the risk of the selected portfolio of US shares, especially to prevent massive loses during unfavourable situation on stock market and otherwise to maximize the profit on bull market. The aim is not to predict which share would be the best to buy, but to predict the sector which would be the best to hold and in what proportion of the entire portfolio.As we can experiencein recent days a coronavirus crisis, we will try to learn model from past crisis and then try to evaluate, if the model managed to avoid massive loses during crisis.For this purpose was selected US stock market because there is the longest series of dataavailable. For the realization of the prediction analysis was selected statistical method of logistic regression and the more advanced methods like decision trees, random forest and neural nets. The suitable programming language for prediction analysis isPython, therefore, it was also used in the analyses presented in the article.