Abstract
We investigate the impact of agent communication networks on prices in an artificial stock market. Networks with different centralization measures are tested for their effect on the volatility of prices. Trading strategies diffuse through the different network topologies, mimetic contagion arises through the adaptive behavior of the heterogeneous agents. Short trends may trigger cascades of buy and sell orders due to increased diffusion speed within highly centralized communication networks. Simulation results suggest a correlation between the network centralization measures and the volatility of the resulting stock prices.