Multiobjective portfolio optimization including sentiment analysis

Abstract

Volatility (or risk) in stock market is a crucial factor that has always been of great interest to investors to facilitate the decision making about their investments. The two core objectives of investors are optimization of volatility and generation of returns at the same time. One can also assume that news can be a factor which can determine volatility when combined with daily returns. In this study we used multiobjective optimization and sentiment analysis of news data together to create two models. In the first multiobjective optimization model, we optimize risk and returns using the conventional formulation and daily returns data. In the second multiobjective optimization model, we again optimize risk and returns but calculate returns differently using daily returns as well as sentiment analysis using news data to see if the model including news behaves differently as compared to the conventional model. The results of both the models have been analyzed in this study. It has been found that while keeping several factors constant, we found no difference in the risk and return of both the models.