Risk-aware two-stage bidding and incentive-compatible allocation for virtual power plants using probabilistic tradable capacity

The increasing integration of distributed energy resources (DERs) introduces significant uncertainty for Virtual Power Plants (VPPs) in electricity spot markets, leading to market risks and potential deviation penalties. This paper presents a unified framework that addresses these challenges by forecasting probabilistic tradable capacity of VPPs using a Bayesian Patch Temporal Shift Transformer (Bayes-PatchTST) to capture the dynamic flexibility of wind, solar, energy storage, and electric vehicle resources; Implementing a two-stage risk-aware bidding strategy for day-ahead and intra-day markets incorporating Conditional Value at Risk (CVaR) to explicitly control market risk; Introducing a hybrid benefit allocation mechanism combining Shapley value and Nash bargaining to ensure incentive-compatible revenue distribution among heterogeneous resources. Case studies demonstrate that the proposed method improves market revenue, reduces deviation penalties, and enhances internal coalition stability, providing a practical and systematic approach for translating physical uncertainty into economically optimized and fair market participation.