This paper discusses the agent-based modeling method from the perspective of complex economics, pointing out the shortcomings of traditional economic theories in explaining real-world economic phenomena and predicting crises. Complex economics emphasizes the non-equilibrium, non-linear, and evolutionary characteristics of economic systems, contrasting with neoclassical economics. Agent-based modeling (ABM) studies the overall behavior of a system by simulating the behavior of individual agents, featuring heterogeneity, self-organization, and other characteristics, making it more realistic in reflecting economic systems. Despite facing challenges in model construction, calibration, and universality, ABM has potential in economic modeling, and with the advancement of data and computing capabilities, it is expected to solve existing problems and promote the development of economic research.