Abstract: With the proliferation of electric vehicles and advancements in vehicle-to-grid interaction technologies, the randomness and complexity of distributed energy transactions are increasingly pronounced. Conventional V2G trading mechanisms typically employ rolling matching schemes with fixed time intervals, which lack sufficient flexibility in responding to real-time fluctuations in market activity. To address this issue, this paper proposes an adaptive rolling-window blockchain-based V2G transaction matching method that incorporates market activity metrics. Market activity is quantified by monitoring two key indicators: the number of unmatched orders and the order growth rate in real time. A dynamic window adjustment algorithm is designed to achieve adaptive scaling of the matching window. Simulation results demonstrate that, compared to fixed-window strategies, the proposed dynamic window algorithm significantly improves transaction success rates and total traded electricity volume while maintaining low average transaction latency. This provides an engineering-feasible solution for optimising V2G trading mechanisms.