A multiagents intermediary-driven combinatorial electronic market model is given. Applying multiagents coalition formation theory for it, a mathematical analysis methodology is used for customer coalition formation. At the macroscopic level of the marketplace, the combinatorial marketplace is described as a stochastic swarm system and the dynamics of multiagents coalition formation is studied. Based on the rate equation in swarm intelligence, a differential equation is derived, which gives the quantitative characteristic for coalition formation process of large-scale customers in the combinatorial market. Simulation experiments are performed to study the feasibility and validity of multiagents coalition formation in combinatorial trading. The global behavior of the system compared with correlative works proves the effective improvement of the dynamic process.