Classical Decomposition of Markowitz Portfolio Selection
Olivia Di Matteo and
Staff Working Papers from Bank of Canada
In this study, we enhance Markowitz portfolio selection with graph theory for the analysis of two portfolios composed of either EU or US assets. Using a threshold-based decomposition of their respective covariance matrices, we perturb the level of risk in each portfolio and build the corresponding sets of graphs. We show that the “superimposition” of all graphs in a set allows for the (re)construction of the efficient frontiers. We also identify a relationship between the Sharpe ratio (SR) of a given portfolio and the topology of the corresponding network of assets. More specifically, we suggest SR = f(topology) ≈ f(ECC/BC), where ECC is the eccentricity and BC is the betweenness centrality averaged over all nodes in the network. At each threshold, the structural analysis of the correlated networks provides unique insights into the relationships between assets, agencies, risks, returns and cash flows. We observe that the best threshold or best graph representation corresponds to the portfolio with the highest Sharpe ratio. We also show that simulated annealing performs better than a gradient-based solver.
Keywords: Central; bank; research (search for similar items in EconPapers)
JEL-codes: C02 (search for similar items in EconPapers)
Pages: 33 pages
New Economics Papers: this item is included in nep-cmp, nep-ore and nep-rmg
References: Add references at CitEc
Citations: Track citations by RSS feed
Downloads: (external link)
https://www.bankofcanada.ca/wp-content/uploads/2020/06/swp2020-21.pdf Full text (application/pdf)
This item may be available elsewhere in EconPapers: Search for items with the same title.
Export reference: BibTeX
RIS (EndNote, ProCite, RefMan)
Persistent link: https://EconPapers.repec.org/RePEc:bca:bocawp:20-21
Access Statistics for this paper
More papers in Staff Working Papers from Bank of Canada 234 Wellington Street, Ottawa, Ontario, K1A 0G9, Canada. Contact information at EDIRC.
Bibliographic data for series maintained by ().