The liberalization of the electricity supply industry has shifted the analyses and modelling activities from planning to operation. However, project studies and investment appraisal still require medium and long-term anticipation of the electricity market prices. The models traditionally used for making such projections i.e. statistical extrapolation or econometrics fail to capture the future structural changes in the emerging electricity markets. There is a need for a novel framework of modelling that could extend game theoretical assumptions to more complex ones. This paper proposes, in a decision-making perspective, a new multi-agent architecture specifically designed to support flexible planning activities in decentralized electricity markets. In this model, the concept of synthetic agents is used for modelling in flexible forms multi-functional market players, possible mergers and coalitions in the electricity market.