We address the problem of finding optimal medium- term generation policies for a specific generation company by modeling the supply side of a liberalized electricity market. The model assumes a noncooperative oligopoly and determines the joint optimal generation policies of all market generators, taking into account hydro, market, and system uncertainties. We propose an endogenous function of market price with respect to load duration where the choice of fuel and technology influences both the average and range of variation of the medium-term market price. We assume an inelastic demand represented by the load-duration curve, which is matched using the Bloom and Gallant formulation. This accounts for unit outages without using scenarios, which are reserved for modeling other uncertainties such as a latent price variable and the hydro inflows. The equilibrium is solved using the Nikaido-Isoda relaxation algorithm, which enables a series of multistage cubic stochastic programming models to be solved. In order to deal with the large number of load matching constraints, we use a heuristic which allows us to generate only those constraints that will presumably be active at the optimal solution. The model is calibrated to the Spanish electricity market using historical price and generation data.