Empirical analyses on portfolio optimization strategies

Abstract

In this study, the success of the application of control engineering approaches to the nancial portfolio construction problem are investigated empirically. Often, the aim of the investor is to maximize the returns while keeping the risks at minimum possible level. In our work, the investor's problem is formulated mathematically by using Modern Portfolio Theory. The risk and return parameters in the formulation are estimated by using historical data. The investment strategy of an investor is characterized by three variables. The rst of these variables is the risk tolerance of the investor. In our work, the investors are divided into three groups according to their risk appetites: Risk Avoiders, Controlled Risk Takers and Adventurers. The second variable is named as the investment horizon. The investment horizon denotes the future time at which the investor hopes to get a return. The third variable determines the length of the historical data that is used for the estimation of the risk and return parameters. By changing the values of these variables inside a nested for loop structure, various investment scenarios are simulated to decide on the most useful investment strategy. The empirically best investment strategy is then applied onto a dynamically updated portfolio and the performance of the portfolio is compared to a speci ed benchmark. In all analyses and simulations, the data from Istanbul Stock Exchange is used. The ISE30 index is chosen as the benchmark. A comprehensive software program with a graphical user interface is developed using MATLAB for simulations and analyses.