The emergence of network communities by the action of coevolving market agents

The agent-based multileader model of the stock price dynamics on the directed evolving complex network is studied by direct simulation. The resulting stationary regime follows from the balance of extremal dynamics, adaptivity of the strategic variables, and reconnection rules. For given parametric combination, the network displays small-world phenomenon with high clustering coefficients and power-law node degree distribution. The fitness exploration by the mechanism of repeated random walk is used to violate dominance of centralized leadership. The simulation suggests that ultra slow dynamics of fitness implies explanation of the long-time volatility of the log-price returns. The text was submitted by the authors in English.