Modelling financial time series using nonlinear and nonparametric bayesian methods

Abstract

The scope of the thesis is to provide an original contribution in modeling financial time series (i.e financial returns). We propose/extend univariate and multivariate models in order to better explain the financial returns and capture the well known stylized facts which characterize them. In particular we propose a Threshold Normal Mixture Garch model, in order to model the fat tails and estimate more efficiently the tail risk. We develop a Threshold Regression Model to examine/identify the nonlinear risk exposures in Hedge fund strategies and to test the existence of market timing abilities. Moreover we propose a new Multivariate Garch Model which allows the different series (i.e Hedge funds) to be affected by different predictors (risk factors). For all these models we consider the problem of estimation, model selection and prediction by using Markov Chain Monte Carlo methods and Bayesian techniques.Moreover we use MCMC methods in the context of Bayesian nonparametric inference in o ...
show more

All items in National Archive of Phd theses are protected by copyright.

Handle URL
http://hdl.handle.net/10442/hedi/29930
ND
29930
Alternative title
Μοντελοποίηση χρηματοοικονομικών χρονολογικών σειρών με τη χρήση μη γραμμικών και μη παραμετρικών μπεϋζιανών μεθόδων
Author
Giannikis, Dimitrios of Sotirios
Date
2013
Degree Grantor
Athens University Economics and Business (AUEB)
Committee members
Βρόντος Ιωάννης
Δελλαπόρτας Πέτρος
Πάνας Επαμεινώνδας
Galeano Pedro
Πολίτης Δημήτριος
Τζαβάλης Ηλίας
Tarantola Claudia
Discipline
Social Sciences
Economics and Business
Keywords
Bayesian inference; Model Selection; MCMC; Time varying volatility; Threshold regression models; Time series; Econometrics; Hedge funds
Country
Greece
Language
English
Description
xxvii, 233 σ., tbls., ch., ind.
Usage statistics
VIEWS
Concern the unique Ph.D. Thesis' views.
Source: Google Analytics.
ONLINE READER
Concern the online reader's opening.
Source: Google Analytics.
DOWNLOADS
Concern all downloads of this Ph.D. Thesis' digital file.
Source: National Archive of Ph.D. Theses.
USERS
Concern the unique Ph.D. Thesis' views.
Source: National Archive of Ph.D. Theses.