Nearly-optimal asset allocation in hybrid stock investment models. (Q703185)

The authors have developed nearly-optimal asset allocation strategies for a class of hybrid stock-investment models that include continuous dynamics and discrete events. The aim of the paper is to reduce the computational complexity based on appropriate aggregation/decomposition and averaging technics, since one needs to distinguish short-term and long-term investors and/or incorporate various economic factors into the model leading to large state space of the Markov chain. Such Markov are considered that have large state spaces with complex structures, which arise due to the inclusion of various economic factors such as market trends, interest rates, and business cycles. The averaging approach is used. The idea is to aggregate the states according to their jump rates and replace the actual system with its average. In the process of obtaining the near optimality, weak convergence results are also derived of a singularly-perturbed switching diffusion as a by-product.