Complexity behaviours of agent-based financial dynamics by hetero-distance contact process

To model the nonlinear and complex dynamics of financial systems, a new model for the formation of financial prices is developed, taking into account heterogeneity in the communication range of market agents. Specifically, one type of agents can potentially gather and disseminate information via additional long-distance contacts compared to the other type, and interactions among these agents are imitated by the contact process. The financial price series of the model are simulated, analysed, and compared with multiple major stock indices in nonlinear fluctuation behaviours. To better investigate the complexity structure of the financial time series, a generalization of the multiscale entropy method is developed to consider various moments in coarse graining. Overall, the modelled series are found to follow a fat-tail distribution and a pattern of complexity structure over both moments and time scales similar to real market data. This similarity is also shown by applying alternative complexity measure, matching energy method. Moreover, the wealth inequality among agents is found to increase over time within each type as well as across two types, further revealing nonlinear price and welfare dynamics of the model.