國立高雄大學圖資館 |

語系: 繁體中文

說明(常見問題)

圖資館首頁

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

A state-space model of financial tim...

Columbia University.

A state-space model of financial time series consistent with technical trading rules.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	A state-space model of financial time series consistent with technical trading rules.
作者:	Nimeskern, Olivier.
面頁冊數:	192 p.
附註:	Adviser: Christopher C. Heyde.
附註:	Source: Dissertation Abstracts International, Volume: 67-10, Section: B, page: 5828.
Contained By:	Dissertation Abstracts International67-10B.
標題:	Statistics.
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3237302
ISBN:	9780542916915

A state-space model of financial time series consistent with technical trading rules.
Nimeskern, Olivier.

A state-space model of financial time series consistent with technical trading rules. - 192 p.

Adviser: Christopher C. Heyde.

Thesis (Ph.D.)--Columbia University, 2006.

The aim of this thesis is to propose a new model of financial time series trends. The motivation is two-fold. First, despite the large number of models introduced in the literature, it has been shown that none is compatible with some results derived from technical trading rules. Additionally, academic research has predominantly focused on variance models and much less on trend models. Second, there is a deep dichotomy between the prediction models used by academia and the ones used by market practitioners, in particular by so-called "technical analysts". As both fields have their own merits, strong theoretical mathematical foundations versus sharp domain knowledge, our model attempts to build on results and stylized facts from both sides. In particular, it rests on the assumption from applied finance that financial time series are locally stationary, changing regimes infrequently, each regime mainly characterized by its drift or trend. This "trend" model is framed under standard academic modeling methods, in particular using a Bayesian hierarchical structure.

ISBN: 9780542916915Subjects--Topical Terms:

182057
Statistics.

A state-space model of financial time series consistent with technical trading rules.
LDR:03219nmm _2200277 _450 001 180694
005 20080111103820.5
008 090528s2006 eng d
020 $a 9780542916915
035 $a 00311720
040 $a UMI $c UMI
100 0 $a Nimeskern, Olivier. $3 264273
245 1 2 $a A state-space model of financial time series consistent with technical trading rules.
300 $a 192 p.
500 $a Adviser: Christopher C. Heyde.
500 $a Source: Dissertation Abstracts International, Volume: 67-10, Section: B, page: 5828.
502 $a Thesis (Ph.D.)--Columbia University, 2006.
520 # $a The aim of this thesis is to propose a new model of financial time series trends. The motivation is two-fold. First, despite the large number of models introduced in the literature, it has been shown that none is compatible with some results derived from technical trading rules. Additionally, academic research has predominantly focused on variance models and much less on trend models. Second, there is a deep dichotomy between the prediction models used by academia and the ones used by market practitioners, in particular by so-called "technical analysts". As both fields have their own merits, strong theoretical mathematical foundations versus sharp domain knowledge, our model attempts to build on results and stylized facts from both sides. In particular, it rests on the assumption from applied finance that financial time series are locally stationary, changing regimes infrequently, each regime mainly characterized by its drift or trend. This "trend" model is framed under standard academic modeling methods, in particular using a Bayesian hierarchical structure.
520 # $a This comprehensive study is performed gradually in order to formulate the model rigorously. After reviewing the background of financial time series models and technical trading rules research, we introduce our data set of fifty financial time series representing disparate asset classes, and run standard statistical tests in order to appreciate some of their distinctive properties. Then, we analyze the open problem of finding the unknown number of change points of a time series assumed to be piecewise stationary. An exact dynamic programming algorithm and an approximate segmentation algorithm with good properties are analyzed and applied to the data set. This enables us to investigate the empirical distribution of the parameters of our hierarchical model, and hence to propose distributional forms for the hyper-parameters and the residuals. The model is then tested using the bootstrap methodology applied to technical trading rules. We conclude by comparing the model goodness of fit versus the main models from academia, and by assessing if it is effective in modeling financial time series and in addressing the efficient market hypothesis.
590 $a School code: 0054.
650 # 0 $a Statistics. $3 182057
650 # 0 $a Economics, Finance. $3 212585
690 $a 0463
690 $a 0508
710 0 # $a Columbia University. $3 212899
773 0 # $g 67-10B. $t Dissertation Abstracts International
790 $a 0054
790 1 0 $a Heyde, Christopher C., $e advisor
791 $a Ph.D.
792 $a 2006
856 4 0 $u http://libsw.nuk.edu.tw:81/login?url=http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3237302 $z http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3237302