This methodological and normative thesis extends the modern economic valuation theory of projects under uncertainty, known as real options theory, from the point of view of an oil company that optimizes the allocation of investment and resources. The real options theory is combined with other theories - so the name hybrid real options - in order to perform a more comprehensive and realistic analysis of complex problems that arises from petroleum industry. The two main combinations analyzed here are: (a) the combination of real options theory with game theory - real options games - to consider endogenously the strategic behavior of other firms, especially in the optimal stopping game with positive externalities known as war of attrition, as well as the possibility to change this game by a cooperative bargain game; and (b) the combination of real options theory with methods from probability theory and Bayesian statistical decision - Bayesian real options - generating a new way to model technical uncertainty of a project in dynamic real options models. These two combinations are re-combined in order to obtain an adequate solution that captures the value of information differences in non-cooperative and cooperative games. Important variables like exploratory chance factor, volume, and quality of a petroleum reserve, are modeled with the development of a new theory on revelation distribution and measures of learning. In a more concise way, are analyzed other hybrid real options, highlighting the combination of real options theory with the evolutionary computation theory - evolutionary real options - with great potential in complex applications of optimization under uncertainty. This method is exemplified with an application using the genetic algorithms to evolve the decision rule for optimal exercise of a real option.