In view of the impact of the volatility of wind power and the randomness of electric vehicle usage on the normal operation of the electricity market, a joint transaction model of the day-ahead market and the real-time market that comprehensively considers "source-network-load--storage" is proposed. The day-ahead market establishes a master-slave game model which considers the configurability of electric vehicles between generators and three types of users, thus forming a forecasted transaction combination between generators and users. Based on the actual power fluctuations on both wind power producers and users, the real-time market aims to minimize social costs and establishes a demand response model that considers the actual power fluctuations on both sides of supply and demand. In order to alleviate the adverse effects of double-sided power fluctuations on the power grid and avoid speculation, a new type of predictive bias penalty mechanism that accounts for both supply and demand fluctuations in power consumption is proposed. Finally, a simulation verifies the feasibility and effectiveness of the constructed model. This work is supported by Science and Technology Major Special Fund of Shanxi Province (No. 20181102028), National Natural Science Foundation of China (No. 51807129), and Key Innovation Team Construction Program of Shanxi “1331” Project (No. 1331KIRT).