Satisfying convex risk limits by trading (Q2488474)

This paper deals with a market with a semimartingale price process. The authors take as given a finite set of valuations and stress measures which are called scenario measures. With each scenario measure a special floor is associated. A random variable, representing the final position of a trading strategy, is deemed acceptable if under each of a variety of probability measures its expectation dominates a floor associated with the measure. The central result of the paper is the characterization of random variables representing the wealth of an agent at a time prior to the final time from which the agent can trade to final acceptability. The representation is simple and shows, in particular, that if one cannot form a martingale measure as a convex combination of scenario measures, then the final acceptability condition imposes no constraint on the initial wealth. The set of initial positions from which the final acceptability can be achieved is a closed half-line. Primal and dual linear programs for computing the bound of this half-line are provided. An application of these ideas to derivative security pricing is developed. Buyer's and seller's prices for contingent claims are defined. These are like utility indifference prices, except they are based on a concept of ``risk-measure indifference''. A stochastic volatility model in which the scenario measures correspond to different levels to which the volatility reverts, is presented. A fairly explicit representation for strategies that trade to acceptability is provided.