Portfolio optimization using higher moments in an uncertain random environment

In this paper, a multi-objective portfolio optimization problem is studied in an uncertain random environment using higher moments. We consider a scenario involving an asset universe wherein some assets have sufficient historical return data for modelling as random variables, and others, listed rela...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Information sciences Jg. 567; S. 348 - 374
Hauptverfasser: Mehlawat, Mukesh Kumar, Gupta, Pankaj, Khan, Ahmad Zaman
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Elsevier Inc 01.08.2021
Schlagworte:
Chance distribution
Multi-objective programming
Portfolio optimization
Real-coded genetic algorithm
Uncertain measure
Uncertain random variable
Portfolio optimization
Chance distribution
Multi-objective programming
Uncertain random variable
Real-coded genetic algorithm
Uncertain measure
ISSN:0020-0255, 1872-6291
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:In this paper, a multi-objective portfolio optimization problem is studied in an uncertain random environment using higher moments. We consider a scenario involving an asset universe wherein some assets have sufficient historical return data for modelling as random variables, and others, listed relatively recently, lack historical data. The asset returns of such assets are modelled as uncertain variables. Thus, a hybrid environment involving both uncertain and random variables is considered. We use mean absolute semi-deviation as a risk measure and employ skewness (i.e., third moment) in the portfolio optimization model. The expressions for the mean absolute semi-deviation and skewness of an uncertain random variable have been derived. We show that the derivation of mean absolute semi-deviation is not based on any stipulation and thus is an accurate measure of risk (unlike the variance of an uncertain random variable, which uses stipulation). We propose a hybrid genetic algorithm as a solution methodology and provide empirical proofs to illustrate its advantages. The proposed methodology has been applied to a case study involving 100 assets listed in the NASDAQ-100 index of the U.S. stock market. We do an ex-post analysis to track the performance of our model out-of-sample and illustrate its advantages.
ISSN:0020-0255
1872-6291
DOI:10.1016/j.ins.2021.03.019