Wirth, Philipp (2025) Graph and tensor models for correlation network-based portfolio construction. PhD thesis, University College London, UK.
Abstract
This thesis develops and evaluates novel methodologies from graph theory and higher-order tensor decomposition to enhance portfolio diversification by uncovering hidden market structures and higher-dimensional asset relationships. The Dynamic Modularity-Spectral Algorithm (DynMSA) is introduced, a graph-based method combining community detection, spectral clustering and Random Matrix Theory to dynamically cluster assets without predefined cluster parameters. Tucker and Tensor Train decompositions are utilised to identify latent, multidimensional factors influencing asset returns over time. Tucker decomposition is applied to the Fama-French 5-factor betas of US stocks to investigate any higher-order relationships in these factor influences on US stock prices. After that, Tensor Train decomposition is used on time-wise stacked correlation matrices of US stocks to find la- tent factors that drive the changes in correlation network structures. Empirical analyses reveal that both DynMSA and Tensor Train decomposition methods generate statistically significant outperformance in portfolio returns and effectively uncover hidden market microstructures, confirming their practical relevance for risk management and portfolio optimisation. While similar performance and factor relationship insights were not consistently observed when decomposing Fama-French factor sensitivities using Tucker Decomposition, the general results of this work highlight the capabilities of non-traditional method- ologies for portfolio construction.
| Item Type: | Thesis (Doctoral) |
|---|---|
| Index terms: | modularity, correlation matrice, methodology, graph theory, clustering, diversification, risk management |
| Subjects: | research methods, theoretical framing, market analysis, design flexibility, risk assessment, data science |
| Topics: | Risk Management, Research Practice, Digital Applications, Design Practice |
| Descriptive scope: | 4 PCTA |
N.B. Descriptive scope is a count of how many of the five facets of empirical research are indicated by the words used in title, abstract and keywords. It is not intended as a judgement on the research; merely a count of the kind of word we would expect to indicate Phenomenon, Concepts, Theoretical framing, Empirical techniques, Analytical techniques. If all five are present, then a code of “5 PCTEA” will indicate this. If you feel the coding for this record is questionable, we welcome discussion around the terms we matched or the way we categorized them. The facet you would expect may not be coded, or a facet may be coded inappropriately. This can also bear on a larger question, of which facets should be treated as defining in construction management research. Please get in touch, and we will look at it. More details here