Self-similarity and network perspective of the Chinese fund market

By testing 88 different funds of the Chinese fund market (CFM), we find fractal behavior and long-range correlations in the return series, which are insensitive to the kind of funds. Meanwhile, a power-law relationship between the deviation D of prices and the Hurst exponent H has been obtained, which may be useful for predicting the price time series. In addition, with funds being viewed as nodes, and the connections among the funds being determined by the cross-correlation coefficients, using a winner-takes-all approach, we investigate the topological properties of the fund network. Our analysis reveals that, during different time periods, the cumulative degree distributions of the fund network all obey the double power-law format. Moreover, the small-world property is also found for the fund network.