Hierarchical topological clustering learns stock market sectors

Doherty, K., Adams, R.G., Davey, N. and Pensuwon, W. (2005) Hierarchical topological clustering learns stock market sectors. In: Procs of 2005 ICSC Congress on Computational Intelligence Methods and Applications :. Institute of Electrical and Electronics Engineers (IEEE). ISBN 1-4244-0020-1
Copy

The breakdown of financial markets into sectors provides an intuitive classification for groups of companies. The allocation of a company to a sector is an expert task, in which the company is classified by the activity that most closely describes the nature of the company's business. Individual share price movement is dependent upon many factors, but there is an expectation for shares within a market sector to move broadly together. We are interested in discovering if share closing prices do move together, and whether groups of shares that do move together are identifiable in terms of industrial activity. Using TreeGNG, a hierarchical clustering algorithm, on a time series of share closing prices, we have identified groups of companies that cluster into clearly identifiable groups. These clusters compare favourably to a globally accepted sector classification scheme, and in our opinion, our method identifies sector structure clearer than a statistical agglomerative hierarchical clustering method


picture_as_pdf
900869.pdf

View Download

Atom BibTeX OpenURL ContextObject in Span OpenURL ContextObject Dublin Core MPEG-21 DIDL Data Cite XML EndNote HTML Citation METS MODS RIOXX2 XML Reference Manager Refer ASCII Citation
Export

Downloads