It has been assumed for quite some time that assets returns are normally distributed, making many portfolio optimization models based on normal distributions. One of the most popular is Markowitz’s portfolio optimization model. The years following the 2008 financial crisis have shown that the technical
progress of financial markets and their globalization have also brought up some new challenges as well as some issues. One of these is the need for a diversification strategy that takes into account large losses and the growing dependence of returns on assets in crisis periods. This also increased the importance of non-Gaussian models and tail dependence in portfolio optimization.
Due to empirical evidence based on various data, it has been generally accepted today by the experts that the values of financial assets indicate a heavy-tailed distribution. The aim of this diploma thesis is to describe the optimization of portfolios, whose values are random variables with heavy tails.
I will present in more detail the classic Markowitz model, the equally weighted portfolio and the optimization strategy based on the extreme risk index. The last method comes from extreme value theory and is especially strong in the case of heavy tails, for which this method is designed. It reduces the likelihood of large losses and can therefore contribute to improved portfolio values, even in times of high market risk.
Finally, I will explore the potential of described optimization methods in practice with the comparison based on backtesting, where I will take the daily share prices of the components of the EURO STOXX 50 index for the period 2001-2011.
|