Autoregressive hidden Markov switching\\models of count data
This paper introduces an alternative hidden Markov switching model. In particular, an autoregressive hidden Markov switching model of count data is formulated and applied to financial data. Through application of this new model, a theoretically-motivated representation of the dynamics of the number of orders placed per unit of time (referred to as order-flow) on the London Stock Exchange is provided. Using the economic arguments of Rock (1996), the suitability of a 2-state autoregressive hidden Markov switching model is demonstrated in this context. This model provides the best fit amongst competing discrete-valued time series models. Moreover, the parameters of this model are found to vary in a predictable manner according to whether the morning, lunch time, or afternoon trading sessions are considered.
|Date of creation:||29 Aug 2002|
|Date of revision:|
|Contact details of provider:|| Postal: Office of the Secretary-General, Rm E35, The Bute Building, Westburn Lane, St Andrews, KY16 9AR, UK|
Phone: +44 1334 462479
Web page: http://www.res.org.uk/society/annualconf.asp
More information through EDIRC
When requesting a correction, please mention this item's handle: RePEc:ecj:ac2002:174. See general information about how to correct material in RePEc.
For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: (Christopher F. Baum)
If references are entirely missing, you can add them using this form.