Abstract: In this talk, I will introduce new notions of conditional random sets that I have developed over the past few years. These concepts provide a powerful framework for addressing problems in mathematical finance, particularly in the context of super-hedging under market incompleteness. My main motivation arises from the need to handle financial market models with transaction costs, where the classical notion of a martingale measure is no longer valid. This setting calls for a novel theoretical approach as well as the development of efficient numerical procedures. I will present several models that have been completely solved within this framework and illustrate that the proposed numerical methods perform effectively on real financial data.