In decision under uncertainty problems that depend on multi- period preferences, as the case of trading electricity contracts in Brazil, agents should expresses their inter and intratemporal preferences. The classical economical approach is to define a time separable von Neumann and Morgenstern utility functional. This functional is composed by the sum of the expected utility of each period times an impatience factor that should express the agent`s intertemporal preference. This approach demands the specification of a subjective impatience factor, which should weight utilities units. Due to this subjectiveness and its estimation difficulties, the applied financial groups started to develop more pragmatic approaches based on risk control. In this sense, companies that maximize expected profit will impose constraints on acceptable losses using estimated occurrence probabilities of different outcomes. In this sense, the economical and applied financial approaches have been diverging in the last four decades and, during this time, many studies and developments have been done in the risk metrics field to generate and prove stability and coherence properties for the proposed metrics. This thesis has three main objectives: (i) propose an alternative approach for multi-period decisions problems based on the present value of the certainty equivalent of each period; (ii) show how this approach can be modeled as a mixed integer linear programming problem (MILP) when adopting a piecewise linear utility function; and (iii) provide connections between utility theory and expected maximization problems constrained to alpha-CVaR risk metrics.