This paper constructs a micro stock market model based on the information propagation and herd behaviors among investors. Unlike present studies, it considers the influence of investor network density on information propagation. By numerical simulation, we conclude some results about how the different relative importance between network density and price fluctuation for the information propagation speed impact the macro stock statistical characteristics. We find out that the relative importance has nothing to do with some basic statistical characteristics, such as volatility clustering, sharp peaks and fat-tail distributions of stock yield. Both the investor clusters and stock absolute yield obey negative power law. The change of relative importance influences fractal feature of stock yields. The certain extent impact of price fluctuation on information propagation speed is a sufficient and necessary condition for state persistence of stock yield.