Agent-based modeling for blockchain-based Ponzi scheme evolution and dynamics

Source abstract

Abstract Blockchain-based Ponzi schemes leverage blockchain’s decentralization and immutability to lure investors and trigger substantial financial losses. This study constructs an agent-based model calibrated with real PlusToken transaction data to examine the evolutionary mechanisms of Blockchain-based Ponzi schemes and quantify how heterogeneous investor behaviors influence the evolution of these schemes. We categorize agents into risk-averse, risk-neutral, and risk-seeking types, each optimizing expected utility based on its historical trading record. The Herfindahl-Hirschman Index and Normalized Shannon Entropy are applied to measure the token concentration among investors. Simulation outcomes show that the fraudster escape scenario yields the fastest collapse and the most severe token concentration, while timely downward adjustments to investor risk tolerance effectively mitigate post-collapse wealth inequality. Additionally, risk-seeking investors display a stronger willingness to join speculative Ponzi schemes yet incur heavier cumulative losses upon the collapse of the scheme.