Browse DORAS
Browse Theses
Search
Latest Additions
Creative Commons License
Except where otherwise noted, content on this site is licensed for use under a:

Stability-based, random matrix theory filtering of financial portfolios

Daly, Justin (2009) Stability-based, random matrix theory filtering of financial portfolios. PhD thesis, Dublin City University.

Full text available as:

[img]PDF - Archive staff only - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
2800Kb
[img]PDF (3rd party copyright material has been removed) - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
2148Kb

Abstract

This thesis describes research on filtering methods using RandomMatrix Theory (RMT) Models in financial markets. In particular, a novel, stability-based RMT filter is proposed and its potential, for reducing stock portfolio risk, is compared to two well-known alternatives. In terms of performance, the stability-based filter achieved 17.3% overall improvement in risk reduction for equally weighted forecasts, and 49.2% for exponentially weighted. Of the filters investigated, not only did it prove to be the most effective and consistent, for overall risk reduction, but was also shown to reduce the frequency of large risk increases, (which, despite their importance, have attracted little attention in the literature to date). The full frequency distribution of filter effects is studied and a comprehensive test methodology established. Improvements, on previous approaches, include integrated use of bootstrap analysis and out-of-sample testing. RMT filtering was also applied to the foreign exchange market, which contains far fewer assets than a typical stock portfolio. Filters were shown to reduce inherent currency trading risks, despite the small number of assets involved. Once again, our novel filter resulted in the lowest risk for exponentially weighted forecasts, and was most consistent in reducing overall levels, exhibiting also the fewest large risk increases. Finally, and more generally, RMT filter testing and analysis can be used to demonstrate the value of rapid response models, i.e. those reacting quickly to market events. Despite the fact that these utilise very recent data, much information is typically masked by noise. Filtering is shown to be successful in exposing such key underlying features.

Item Type:Thesis (PhD)
Date of Award:November 2009
Refereed:No
Supervisor(s):Ruskin, Heather J. and Crane, Martin
Uncontrolled Keywords:Random Matrix Theory (RMT); portfolio analysis; eigenanalysis; eigenvalues; eigenvectors; stability; covariance; correlation;
Subjects:Mathematics > Mathematical models
Mathematics > Probabilities
Physical Sciences > Statistical physics
Computer Science > Computer simulation
DCU Faculties and Centres:DCU Faculties and Schools > Faculty of Engineering and Computing > School of Computing
Use License:This item is licensed under a Creative Commons Attribution-NonCommercial-No Derivative Works 3.0 License. View License
Funders:Irish Research Council for Science Engineering and Technology
ID Code:14941
Deposited On:17 Nov 2009 15:03 by Martin Crane. Last Modified 17 Nov 2009 15:03

Archive Staff Only: edit this record