Post by Sapphire Capital on Jul 11, 2008 21:49:08 GMT 4
Joint Modeling of Extremes and their Clustering with Applications in Finance
KAM HAMIDIEH
University of Michigan at Ann Arbor - Statistics and Management Science
STILIAN STOEV
Boston University - Department of Mathematics and Statistics
GEORGE MICHAILIDIS
Affiliation Unknown
--------------------------------------------------------------------------------
March 17, 2008
Abstract:
We develop a method for estimating the Value-at-Risk based on modeling the arrival intensity and the sizes of the extremes via the Autoregressive Conditional Duration (ACD) and the Generalized Pareto Distribution (GPD). The method is illustrated and backtested on eight assets. The results show that the proposed method is capable of accounting for both the clustering structure and the magnitude of the extremes better than GARCH and unconditional GPD based methods. Further, the examination of only the estimated intensities provides a novel and a useful method for studying the clustering properties of the extreme values.
papers.ssrn.com/sol3/papers.cfm?abstract_id=1107538
KAM HAMIDIEH
University of Michigan at Ann Arbor - Statistics and Management Science
STILIAN STOEV
Boston University - Department of Mathematics and Statistics
GEORGE MICHAILIDIS
Affiliation Unknown
--------------------------------------------------------------------------------
March 17, 2008
Abstract:
We develop a method for estimating the Value-at-Risk based on modeling the arrival intensity and the sizes of the extremes via the Autoregressive Conditional Duration (ACD) and the Generalized Pareto Distribution (GPD). The method is illustrated and backtested on eight assets. The results show that the proposed method is capable of accounting for both the clustering structure and the magnitude of the extremes better than GARCH and unconditional GPD based methods. Further, the examination of only the estimated intensities provides a novel and a useful method for studying the clustering properties of the extreme values.
papers.ssrn.com/sol3/papers.cfm?abstract_id=1107538