An Optimal Control Approach to Portfolio Optimisation with Conditioning Information

In the classical discrete-time mean-variance context, a method for portfolio optimisation using conditioning information was introduced in 2001 by Ferson and Siegel ([1]). The fact that there are many possible signals that could be used as conditioning information, and a number of empirical studies that suggest measurable relationships between signals and returns, causes this type of portfolio optimisation to be of practical as well as theoretical interest. Ferson and Siegel obtain analytical formulae for the basic unconstrained portfolio optimisation problem. We show how the same problem, in the presence of a riskfree asset and given a single conditioning information time series, may be expressed as a general constrained infinite-horizon optimal control problem which encompasses the results in [1] as a special case. Variants of the problem not amenable to closed-form solutions can then be solved using standard numerical optimal control techniques. We extend the standard finite-horizon optimal control sufficiency and necessity results of the Pontryagin Maximum Principle and the Mangasarian sufficiency theorem to the doubly-infinite horizon case required to cover our formulation in its greatest generality. As an application, we rephrase the previously unsolved constrained-weight variant of the problem in [1] using the optimal control framework and derive the specific necessary conditions applicable. Finally, we carry out simulations involving numerical solution of the resulting optimal control problem to assess the extent to which the use of conditioning information brings about practical improvements in the field of portfolio optimisation.