| dc.description.abstract | Value at Risk (VaR) is widely used risk measure in risk management. It is defined as the maximum probable loss on a given portfolio
under normal circumstances, commonly accepted as a standard measure of market risk. In this thesis we use combining Copula functions,
Extreme Value Theory (EVT) and GARCH models to estimate portfolio. We apply this approach to a portfolio consisting of stock indices
from VNINDEX and NASDAQ. Before estimating VaR, firstly, The
marginal model distribution of each log return series is build on an
asymmetric GARCH model and EVT( Extreme Value Theory ) to
connect the marginal distribution together to take shape a distribution of multivariate by using Copula functions ( Gaussian, Student’s
t, Clayton, Gumbel and Frank ). Afterthat, we apply Monte Carlo
Simulation (MCS) approach to estimates of the portfolio VaR. Finally,
we use Backtesting methods to check the goodness of fit of approach.
From the results, we conclude that GARCH-EVT-Students t Copula
is better than all other GARCH-EVT-Copulas and traditional methods such as Historical Simulation (HS) and Variance Covariance (VC).
Key words: Value at Risk (VaR), Copula, GARCH, Extreme Value
Theory (EVT), Backtesting. | en_US |
