Litinska A. The information technology of hedge-funds portfolio optimization with various distribution of risk factors

The dissertation is devoted to solving the problems of nonlinear assets portfolio optimization, in particular the fund of hedge funds, when a hypothesis of a normal distribution of risk factors is not fulfilled. The modern approaches to portfolio construction are analyzed. Because of the high kurtosis, negative asymmetry of hedge fund's profitability distribution and their nonlinear dependence on the costs of base assets, the author proposes to use the modification of Uryasev and Rokafelar's method in such cases. To solve the "thick tails" problem it was suggested to approximate the time series by the method of scenario generation, with the help of elliptically contoured distributions. New approach of the loss function presentation is developed, which uses the vehicle of quadratic approximation and bases on a hypothesis about the elliptic distribution of risk factors. In order to take into account an investment attractiveness of the portfolio, our model was supplied by such additional components as the rate of discounting and a level of alternative asset's profitability. The loss function is formulated for the case of the information uncertainty by presenting the model elements as fuzzy numbers. The Uryasev and Rokafelar method is improved in part of possible risk values estimation in the problem of portfolio optimization in the case of profitability maximization. It is suggested to use the modification of simplex method with the eliminative form of inversion and special rule for pivot column selection in order to get the decision of the put by the problem. An original decision support system has been developed and implemented in the field of risk management and investment analysis.