This paper proposes a novel conditional value at risk（CVaR） method based on the quantile， and minimizes the weighted sum of the absolute deviation of each CVaR to build a mathematical model. Using the Monte Carlo method， the model proposed is extended by deducing the generation yield expression corresponding to various marketing risks， such as the price risk， the inflow uncertainty and the demand uncertainty， confronted in the course of the on-grid bidding. The model can be applied to both the time and space distributions of power generation with risk constraints. An annual volume regulation of hydropower in a completely competitive market is simulated to show that the risk measurement index proposed is feasible and applicable.