A0479
Title: Which early warning signals predict high-frequency extreme price movements?
Authors: Julien Hambuckers - University of Liege (Belgium)
Philippe Hubner - HEC Liege, University of Liege (Belgium) [presenting]
Abstract: The dynamic distribution of block maxima time-series is modeled, with an application to high-frequency stock returns. To do so, an autoregressive structure is considered in the parameters of the generalized extreme value (GEV) distribution, which are also conditioned by past information. The recently developed penalization technique is used to select relevant covariates. In a simulation study, the finite sample properties of the estimation are inspected, and the ability to select active covariates and related computational issues is addressed. As an empirical illustration, these techniques are applied for distributional forecasting using 5-minute returns on a sample of NASDAQ stocks, with liquidity measures among the candidate covariates.
