E-Locus - Institutional Repository of the University of Crete - Value at risk, forecasting and non-linear tsansfornations

Home Value at risk, forecasting and non-linear tsansfornations

Results - Details

[Add to Basket]

Identifier

000430110

Title

Value at risk, forecasting and non-linear tsansfornations

Alternative Title

Αξία σε κίνδυνο, πρόβλεψη και μη γραμμική μετασχηματισμοί

Author

Τσάκου, Αλεξάνδρα-Μαρία

Thesis advisor

Τσιώτας Γεώργιος

Abstract

The VaR measure is a statistical quantile, expressing the minimum loss of a security or a portfolio for a given period and confidence level. The accuracy of the assessment of this measure is important as it contributes to the optimal portfolio selection and generally speaking, defines the level of risk that should be faced by an investor or by an institution. At this thesis, I am going to use the semi-parametric CAViaR model in order to estimate VaR. The main aim of this study is the comparison of the existing models and forecasting methods of VaR with those that use non-linear transformations. The above comparison would focus on assessing the predictive power to estimate the VaR measure between the initial and transformed data. The empirical part of this paper, is based on the forecasting ability of the above process for three financial indexes, Cac40, Nikkei225 and SP500.

Language

Greek

Subject

CAViaR models

Forecasting and non-linear tsansfornations

Value at risk

Αξία σε κίνδυνο

Μοντέλα CAViaR

Πρόβλεψη μη γραμμική μετασχηματισμοί

Issue date

2017

Collection