ISBN:  Unavailable

Category: Unavailable

Page count: 25

Market risk can be described as potential losses in portfolio value caused by price changes in the investor's portfolio. Value-at-Risk (VaR) quantifies a loss bound that cannot be exceeded with a specified probability at a given time horizon, i.e., a quantile of the portfolio's loss distribution. We cannot determine this distribution of portfolio losses analytically for portfolios with nonlinear loss functions - especially those portfolios that include options - even if the distribution of risk factors is multivariatenormal. In such cases it is common practice to use extensive approximations and simulations under partly restrictive assumptions. To avoid such reductions, this paper uses approaches based on artificial neural networks (ANN).