Abstract (eng)
Conditional Acceptability Mappings quantify the degree of desirability of random variables modeling financial returns, accounting for available, non-trivial information. They are defined as mappings probability spaces, where nontrivial information is available.
Additionally, such mappings have to be concave, translation- equivariant and monotonically increasing.
Based on the order characteristics of Lp-spaces, superdifferentials and concave conjugates for conditional acceptability mappings are defined and analyzed. The novelty of this work is that the almost sure partial order is consequently used for this purpose, which results in simpler definitions and proofs, but also accounts for all requirements concerning continuity, integrability and measurability of the supergradients and conjugates.
Furthermore, the results about conditional mappings are used to show properties of multiperiod acceptability functionals that are based on conditional acceptability mappings, such as SEC-functionals and acceptability compositions. A chain rule for superdifferentials as well as the conjugate of multiperiod functionals and their properties are derived.