Authors

Xiaochu Zhang

Type

Text

Type

Dissertation

Advisor

Mullhaupt, Andrew | Frey, Robert | Rachev, Svetlozar | Holod, Dmytro | Smith, Noah.

Date

2015-08-01

Keywords

Applied mathematics

Department

Department of Applied Mathematics and Statistics.

Language

en_US

Source

This work is sponsored by the Stony Brook University Graduate School in compliance with the requirements for completion of degree.

Identifier

http://hdl.handle.net/11401/76560

Publisher

The Graduate School, Stony Brook University: Stony Brook, NY.

Format

application/pdf

Abstract

In the first part of this dissertation research, an extension of the binomial tree model to a regime-switching volatility model in a two-state setting for volatility is derived, analyzed and tested. A dynamic programming method for mean-variance hedging is applied to price European option value. After convergence and simulation study, we demonstrate that an HMM driven stochastic volatility process will converge to a geometric brownian motion with a constant volatility. In the second part, we further incorporate an autoregressive component into the regime switching model based on observations of the first part and derive an autoregressive regime-switching model for financial time series data. A parsimonious estimation method of autoregressive regime-switching model is developed using Gram-Schmidt orthogonalization, Frobenius norm minimization, and the EM algorithm. The positive semi-definite correlation matrix issue is considered and addressed in our estimation method. Stability and accuracy is also examined in part two. In the third part, observations based on the analysis of real financial time series are shown. The integrated, fractional integrated and heavy tail feature of non-stationary time series is studied. Estimation, forecasting and backtesting are performed with ARMA-GARCH, FARIMA-FIGARCH and our autoregressive regime-switching model. In comparison with other models, autoregressive regime-switching model has better backtesting results for forecasting VaR models with high frequency financial time series. | 119 pages

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.