| Abstract [eng] |
The main goal for the investor is to optimally distribute his wealth between several securities. However, it’s not an easy task, as price of the securities might vary significantly due to many external factors, such as central banks regulation, raw materials price, technological breakthrough, political situation and etc. Therefore, for this purpose investors most commonly use decision-making techniques, that are based on mathematical optimization or heuristic parameters. The problem is that all methods rely on assumptions, that might be valid during one time period, but not valid during another. One of many assumptions states that investors’ decisions are only rational. This assumption is especially common with mathematical optimization methods. The idea behind it is that rational investor will always invest into the portfolio, which judging on the existing information, will allow to earn the best return for the accepted risk. However, researches show that rationality of the investors is more of the exception rather than rule, therefore methods that rely on the opposite assumption are also under the development. The goal of this work is to compare different investment portfolio formation methods, and to create the multicriteria decision making algorithm for portfolio optimization by combining different type methods. For investment portfolio formation we will use CVaR and diversification parameter optimization methods as well as classical Markowitz model with robust covariance matrix estimation techniques. Also, inertia momentum and moving average methods will be used as these methods do not rely on the investor rationality assumption. Additionally, in this work we will create simple investment portfolio formation technique, that will rely on leading economic indicators by Organization for Economic Co-operation and Development. For scenario generation, Student copula function will be used. Also, multicriteria decision making algorithm will be created by combining single criteria methods with automatic estimation of their significance. All methods under the consideration will be compared using various econometric and statistic parameters as well as tests. Results derived in this work show that during the analyzed period from 1988 to 2018 inclusively, leading economic indicators method allowed to earn the highest return if compared to other models tested in this work. Multicriteria model generated 3,8% smaller average annual return. However, multicriteria model is less risky and because of this it has a similar return to risk ratio as a leading economic indicators portfolio. These two methods can be a good alternative for practically used investment portfolio formation techniques. |
