Complex correlation based networks - clustering and causalities in the market

Mateusz J. Wilinski 3Hideaki Aoyama 1Hiroshi Iyetomi 2

1. Graduate School of Science, Kyoto University, Kyoto 606-8502, Japan
2. Niigata University, Ikarashi-2 8050, Niigata 950-2181, Japan
3. Uniwersytet Warszawski, Wydział Fizyki, ul. Pasteura 5, Warszawa 02-093, Poland

Abstract

I would like to introduce a method of calculating complex correlations in the market, using Hilbert transformation of financial data. Based on these correlations, a complete graph of dependencies can be created, where we can not only distinguish the strength of correlations but also the directions, which implies causalities in the market. For this completely new structure, I will propose some new and different methods of analysis. First, applying the Complex Principal Component Analysis and creating a synchronisation network from which the market clustering can be easily derived. Second, using the Edmonds algorithm to find the directed minimal spanning tree that describes both the structure and the causalities observed in the market.

 

Related papers
  1. Modeling Endogenous Contagion on O/N Interbank Market
  2. Intraday correlation structure for high frequency financial data

Presentation: Poster at 8 Ogólnopolskie Sympozjum "Fizyka w Ekonomii i Naukach Społecznych", by Mateusz J. Wilinski
See On-line Journal of 8 Ogólnopolskie Sympozjum "Fizyka w Ekonomii i Naukach Społecznych"

Submitted: 2015-09-08 17:56
Revised:   2015-09-09 09:33