Dynamical local models for segmentation and prediction of financial time series

M Azzouzi, Ian T. Nabney

Research output: Contribution to journalArticle

Abstract

In the analysis and prediction of many real-world time series, the assumption of stationarity is not valid. A special form of non-stationarity, where the underlying generator switches between (approximately) stationary regimes, seems particularly appropriate for financial markets. We introduce a new model which combines a dynamic switching (controlled by a hidden Markov model) and a non-linear dynamical system. We show how to train this hybrid model in a maximum likelihood approach and evaluate its performance on both synthetic and financial data.
Original languageEnglish
Pages (from-to)289-311
Number of pages23
JournalEuropean Journal of Finance
Volume7
Issue number4
DOIs
Publication statusPublished - 2001

Fingerprint

Financial data
Segmentation
Hybrid model
Financial time series
Train
Generator
Prediction
Hidden Markov model
Stationarity
Nonlinear dynamical systems
Maximum likelihood
Nonstationarity
Financial markets

Bibliographical note

This is a preprint of an article submitted for consideration in the European Journal of Finance © 2001 copyright Taylor & Francis; European Journal of Finance is available online at: http://www.informaworld.com/openurl?genre=article&issn=1351-847X&volume=7&issue=4&spage=289

Keywords

  • NCRG

Cite this

Azzouzi, M ; Nabney, Ian T. / Dynamical local models for segmentation and prediction of financial time series. In: European Journal of Finance. 2001 ; Vol. 7, No. 4. pp. 289-311.
@article{4a1c7bf07a354300a48ea027d2351c04,
title = "Dynamical local models for segmentation and prediction of financial time series",
abstract = "In the analysis and prediction of many real-world time series, the assumption of stationarity is not valid. A special form of non-stationarity, where the underlying generator switches between (approximately) stationary regimes, seems particularly appropriate for financial markets. We introduce a new model which combines a dynamic switching (controlled by a hidden Markov model) and a non-linear dynamical system. We show how to train this hybrid model in a maximum likelihood approach and evaluate its performance on both synthetic and financial data.",
keywords = "NCRG",
author = "M Azzouzi and Nabney, {Ian T.}",
note = "This is a preprint of an article submitted for consideration in the European Journal of Finance {\circledC} 2001 copyright Taylor & Francis; European Journal of Finance is available online at: http://www.informaworld.com/openurl?genre=article&issn=1351-847X&volume=7&issue=4&spage=289",
year = "2001",
doi = "10.1080/13518470110071155",
language = "English",
volume = "7",
pages = "289--311",
journal = "European Journal of Finance",
issn = "1351-847X",
publisher = "Routledge",
number = "4",

}

Dynamical local models for segmentation and prediction of financial time series. / Azzouzi, M; Nabney, Ian T.

In: European Journal of Finance, Vol. 7, No. 4, 2001, p. 289-311.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Dynamical local models for segmentation and prediction of financial time series

AU - Azzouzi, M

AU - Nabney, Ian T.

N1 - This is a preprint of an article submitted for consideration in the European Journal of Finance © 2001 copyright Taylor & Francis; European Journal of Finance is available online at: http://www.informaworld.com/openurl?genre=article&issn=1351-847X&volume=7&issue=4&spage=289

PY - 2001

Y1 - 2001

N2 - In the analysis and prediction of many real-world time series, the assumption of stationarity is not valid. A special form of non-stationarity, where the underlying generator switches between (approximately) stationary regimes, seems particularly appropriate for financial markets. We introduce a new model which combines a dynamic switching (controlled by a hidden Markov model) and a non-linear dynamical system. We show how to train this hybrid model in a maximum likelihood approach and evaluate its performance on both synthetic and financial data.

AB - In the analysis and prediction of many real-world time series, the assumption of stationarity is not valid. A special form of non-stationarity, where the underlying generator switches between (approximately) stationary regimes, seems particularly appropriate for financial markets. We introduce a new model which combines a dynamic switching (controlled by a hidden Markov model) and a non-linear dynamical system. We show how to train this hybrid model in a maximum likelihood approach and evaluate its performance on both synthetic and financial data.

KW - NCRG

UR - http://www.informaworld.com/openurl?genre=article&issn=1351-847X&volume=7&issue=4&spage=289

U2 - 10.1080/13518470110071155

DO - 10.1080/13518470110071155

M3 - Article

VL - 7

SP - 289

EP - 311

JO - European Journal of Finance

JF - European Journal of Finance

SN - 1351-847X

IS - 4

ER -