Portfolio Selection: A Novel Method of Measuring Investment Risk

Abstract

This article proposes a novel portfolio selection model by adopting the basic principles of the classical mean-variance portfolio optimization framework. This model introduces a unique approach to measure risk in an investment. It works under the assumption that the investors establish a target return for their portfolio, considering the achievement of this target as a gain and its failure to be met as a loss. Probabilities of such gains and losses are calculated using normal distribution, generating a curve in two-dimensional space representing potential portfolio outcomes. Concurrently, an investor’s risk tolerance is depicted by another curve on the same two-dimensional space. The primary objective of this model is to optimize the portfolio in alignment with the investor’s risk tolerance curve. To achieve diversification, Shannon entropy is employed as one of the constraints in the model. Given an investor’s risk tolerance, multiple optimal portfolios can be constructed, allowing investors to select portfolios according to their preferences. The results demonstrate high levels of accuracy and promise, suggesting that the proposed model effectively manages investment risk within this domain. A series of experiments have been conducted, and comparisons have been drawn with existing models in the literature.