Generation rights trading (GRT) of dam-type hydropower plants (known as large plants) and runoff hydropower plants (known as small plants) is one of the effective ways to improve the hydropower grid access rate and solve the problem of water spillage and excess local power generation. However, how to choose trading partners and allocate annual electricity in a bilateral negotiation is always a problem. Therefore, a bi-level model for trading partners selection and risk evaluation considering the GRT between large plant and small plants is proposed. First, from the income, cost and other factors in the GRT, the indices of large plant’s trading partners can be determined. Then considering the impact of all indicators the best trading partner will be selected. Then, from the impact of a large plant’s bidding price on the trading share, we refer to the market demand function in econometrics. Then the bidding model of a large plant can be established. Finally, the proportion of the optimal contract, spot and GRT quantity can be determined by CVaR and the Sharpe ratio. The method is applied to the cascaded hydropower plants located in southwest China. The results indicate that the proposed model can reduce the spill of small plants, and keep the profit of a large plant at a high level while reducing risk. It becomes of significance for a large plant to participate in GRT and power allocation.