Portfolio Optimization through Forward Looking Approach: an analysis of the COVID19 market crash

Abstract

The main purpose of this dissertation is the comparison between two different approaches, the historical and the forward-looking ones as to improve portfolio optimization results. The idea is that, applying the forward looking approach based on the implied moments derived by the options written on the assets in the portfolio, it is possible to obtain better results in terms of volatility and returns than using the historical data. This approach is applied to the Partical Swarm Optimization (PSO) and an improved version of the PSO, which will be compared appling both the approaches to determine if the covariance matrix built using the forward looking approach is useful to improve the result of the portfolio. Morever, the performances of the improved PSO algorithm will be tested to understand if it can be considered better in both scenarios.

The thesis is structured as follow: in the first chapter an overview of the Market Porfolio Theory, which can be considered as the theoretical foundation of the porfolio selection, will be presented together with its limits and some possible improvements that can be applied. The second chapter will focus on metaheuristic algorithms, introducing the ide behind them, their formulation and a comparison between the two algorithms. The third chapter introduces the forward looking approach along with an introduction to options and options pricing following the Black-Scholes-Merton model. Lastly, the fourth chapter will discuss the results of the tests conducted comparing the algorithms and the different approaches, the historical and the forward-looking one.