Forecast Precision of Value at Risk: An Evaluation of ARCH-Type Models
Sammanfattning: Over the recent years, value at risk has become an industry standard for measuring downside market risk. This thesis aims to give a thorough differentiation between the different types of models used to estimate value at risk. It shows what gains to be achieved going from a symmetric to an asymmetric model as well as a few other tweaks that can be done to potentially improve the estimates. We provide a detailed comparison of several different kinds of approaches coupled with a thorough backtesting procedure giving a precise evaluation of every alternative approach. Our observations show the importance of choosing a model with regards to its end use and the importance of not adhering too much to convention. Moreover, we show that the asymmetric models prove superior in the analysed stock indices though it is hard to predict the winning model beforehand.
HÄR KAN DU HÄMTA UPPSATSEN I FULLTEXT. (följ länken till nästa sida)