An approach to Hang Seng Index in Hong Kong stock market based on network topological statistics

Abstract

Using homogenous partition of coarse graining process, the time series of Hang Seng Index (HSI) in Hong Kong stock market is transformed into discrete symbolic sequences S={S1S2S3...}, Si ∈ (R, r, d, D). Weighted networks of stock market are constructed by vertices that are 16 2-symbol strings (i.e. 16 patterns of HSI variations), and encode stock market relevant information about interconnections and interactions between fluctuation patterns of HSI in networks topology. By means of the measurements of betweenness centrality (BC) in networks, we have at least obtained 3 highest betweenneess centrality uniform vertices in 2 order of magnitude of time subinterval scale, i.e. 18.7% vertices undertake 71.9% betweenness centrality of networks, showing statistical stability. These properties cannot be found in random networks; here vertices almost have identical betweenness centrality. By comparison to random networks, we conclude that Hong Kong stock market, rather than a random system, is statistically stable.