The sensitivity analysis of electricity price is all along one of the important and unsolved problems in electricity market research， despite its significance for the market participants. Because of the complexity of electricity market， many traditional methods may experience difficulties in solving the price sensitivity problems. A novel method for electricity price sensitivity analysis based on numerical simulation is presented. First according to statistical learning theory， price simulation model is built to accomplish the mapping of price and its influence factors by support vector machine（SVM）. Then based on that model， the changes in prices are simulated with market factors changing at the same time. After that， the sensitivity coefficients are computed， by using the simulated price samples in different market scenarios. In this way， price sensitivity analysis is effectively accomplished by combining the black-box modeling of SVM. The proposed model is featured with its theoretical simplicity， explanation ability and generalization capability. Case studies are used to testify the proposed model.