Network of networks: A meta-model for simulated financial markets

Abstract

We investigate the properties of a calibrated network structure of an agentbased model for a simulated financial market. A meta-model of a network of networks is introduced to capture the simulated market structure. The agent-based model consists of heterogeneous agents characterized by two-dimensional attributes that are investment behavior and investment strategy. The resulting groups of agents are viewed as subnetworks giving rise to a network of networks (NoN). The aggregation of activities of agents in a subnetwork trickles up to shape the aggregate activities of the NoN. The objective of introducing the NoN is to provide a testbed for complex models of simulated markets. Furthermore, we investigate the emergence of the market patterns in terms of prices, moments of returns, market capital, and wealth distributions. The investigation was performed for fully connected homogeneous agents. The results show a significant difference in the market emergence behaviors in terms of prices and returns, however, the market capitalization stays close to the calibrated financial market. Also, the deviation of wealth distributions was less than those in the heterogeneous market.