Change search
ReferencesLink to record
Permanent link

Direct link
Algorithmic evaluation of Parameter Estimation for Hidden Markov Models in Finance
KTH, School of Engineering Sciences (SCI), Mathematics (Dept.), Mathematical Statistics.
2014 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

Modeling financial time series is of great importance for being successful within the financial market. Hidden Markov Models is a great way to include the regime shifting nature of financial data. This thesis will focus on getting an in depth knowledge of Hidden Markov Models in general and specifically the parameter estimation of the models. The objective will be to evaluate if and how financial data can be fitted nicely with the model. The subject was requested by Nordea Markets with the purpose of gaining knowledge of HMM’s for an eventual implementation of the theory by their index development group. The research chiefly consists of evaluating the algorithmic behavior of estimating model parameters. HMM’s proved to be a good approach of modeling financial data, since much of the time series had properties that supported a regime shifting approach. The most important factor for an effective algorithm is the number of states, easily explained as the distinguishable clusters of values. The suggested algorithm of continuously modeling financial data is by doing an extensive monthly calculation of starting parameters that are used daily in a less time consuming usage of the EM-algorithm.

Place, publisher, year, edition, pages
TRITA-MAT-E, 2014:12
Keyword [en]
Hidden Markov Models, Parameter Estimation, Expectation Maximization
National Category
Probability Theory and Statistics
URN: urn:nbn:se:kth:diva-141187OAI: diva2:696772
Subject / course
Mathematical Statistics
Educational program
Master of Science - Mathematics
Available from: 2014-02-15 Created: 2014-02-11 Last updated: 2014-02-15Bibliographically approved

Open Access in DiVA

fulltext(2637 kB)83 downloads
File information
File name FULLTEXT01.pdfFile size 2637 kBChecksum SHA-512
Type fulltextMimetype application/pdf

By organisation
Mathematical Statistics
Probability Theory and Statistics

Search outside of DiVA

GoogleGoogle Scholar
Total: 83 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Total: 401 hits
ReferencesLink to record
Permanent link

Direct link