DSpace 久留米大学

Kurume University Institutional Repository >
紀要論文 >
経済学部 >
経済社会研究 >
第57巻第1・2合併号 >

Please use this identifier to cite or link to this item: http://hdl.handle.net/11316/643

Title: Detecting Change Points and Structural Changes in Stock Price Time Series based upon a Bayesian Approach
Authors: 譚, 康融
Tan, Kangrong
Issue Date: 25-Mar-2017
Publisher: 久留米大学経済社会研究会
Abstract: This paper deals with the detection of change points and structural changes in the time series og stock prices. So far, many studies on how to locate change points and detect the structural changes have been carried out, though, the most of them have built their theoretical and technical platforms and methodologies under the circumstances of normality. But when it is applied to the analysis of the behavior of stock market, as well as many quantitative researches have pointed aut, that the normality of the distribution of a stock price or return usually doesn't hold. It turns out to be a bias estimation to the change point problem. Furthermore, it does harm to the practices in real business world, such as risk measurement and management. On the other hand, it is a necessitated to identify exatly the locations of change points if they do exist in the time series data which we are concerned about. As to solve this problem, we propose a new approach to locate where a change point exactly lies at, or where a structural change exactly occurs. Our proposed approach is firstly to consider the change sizes of the prices, and secondly to deal with the counting number of those chage sizes as a Poisson process, and then to detect the change points based upon a Bayesian approach. Through our numerical analyses, including the artificial dataset and real stock daataset, we find out that our proposed approach works well, and it can be applied to those problems with non-Gaussian phenomena.
URI: http://hdl.handle.net/11316/643
ISSN: 2433-2682
Appears in Collections:第57巻第1・2合併号

Files in This Item:

File Description SizeFormat
keisya57_1-2_1-22.pdf953.52 kBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.


Valid XHTML 1.0! DSpace Software Copyright © 2002-2010  Duraspace - Feedback