Research On Properties Of Fund-stock Bipartite Network Based On Fund’s Share Holding

In recent years, the development of China’s financial markets have greatly contributed to thefund market, the number and size of the fund have increased dramatically, and the fund’s investmentstrategy and investment behavior impact the stock market greatly. According to the China BankingRegulatory Commission statistics, in March2013, there were1,257securities investment funds,which accounts for12.13%of the total market value of shares of listed companies of China, thefund market is cause for concern in the current stock market. The fund’s holdings in the stockmarket has influenced the securities market significantly, along with the growing size of the fund,Meanwhile, research the fund’s holdings will help us to understand the fund’s holdings fully.In20th century, scientists began to study the structure and properties of complex networksfrom the small world networks and scale-free networks, which is widely used in productionprocesses, in all fields of social research social network. The development of the complex networktheories has contributed to a better understanding of the intrinsic links and laws of nature andsociety.Complex network theory and method will be a gold mineμ to broaden the traditional theoryof financial economics. In fact, before the introduction of complex network theory and methods inthe field of financial economics research, interdisciplinary researchers have used complex networkanalysis in other areas and had achieved a breakthrough.We build the "Fund-Security" bipartite network based on the open-end fund holdingsinformation in the stock market, in order to explore the fund holding preference and the basiccharacteristics of complex networks.Based on the open-end fund’s share holding from2010to2012, we build fund holdingdynamic binary networks where nodes correspond to stocks and funds and edges correspond toshareholding partnership, through the weight projection method we get the fund’s share holdingweighted dynamic networks and fund weighted dynamic networks, analyze the networks’ basiccharacteristics and community structures to explore the evolution of fund market. The presentnetworks are small-world networks with stable short average path length and high degree ofclustering, but their degree distributions don’t follow a double power law. We detect the communitystructure of the fund’s share holding weighted networks and fund weighted dynamic networks withweighted CNM algorithm and find that the modularity Q is very small which reflected the diversification strategies of funds. We further analyze these communities by stocks’ industryclassification and find that the funds prefer to invest more industries over time.