Locally time homogeneous time series modelling

Mstislav Elagin, Vladimir Spokoiny

Research output: Chapter in Book/Report/Conference proceedingChapterpeer-review

Abstract

Modelling particular features (stylized facts) of financial time series such as volatility clustering, heavy tails, asymmetry, etc. is an important task arising in financial engineering. For instance, attempts to model volatility clustering, i.e. the tendency of volatility jumps to appear in groups followed by periods of stability, led to the development of conditional heteroskedastic (CH) models including ARCH by Engle (1982) and GARCH by Bollerslev (1986) as well as their derivatives. The main idea underlying the mentioned methods is that volatility clustering can be modelled globally by a stationary process. The chapter is organized as follows. Section 17.2 is devoted to the formulation of the problem and theoretical introduction. Section 17.3 describes the methods under comparison. In Section 17.4 the procedure for obtaining critical values, essential parameters of the procedures, is given. Section 17.5 shows the application of the adaptive methods to the computation of the value-at-risk.

Original languageEnglish
Title of host publicationApplied Quantitative Finance
Subtitle of host publicationSecond Edition
PublisherSpringer Berlin Heidelberg
Pages345-361
Number of pages17
ISBN (Print)9783540691778
DOIs
Publication statusPublished - 2008
Externally publishedYes

Fingerprint

Dive into the research topics of 'Locally time homogeneous time series modelling'. Together they form a unique fingerprint.

Cite this