Parameter & Variable-Selection Uncertainty in Asset Allocation

This book focuses on the problem of finding the optimal allocation strategy in a financial portfolio, using an econometric point of view. Its main contribution is the investigation of a new Bayesian approach for the portfolio choice. Markov Chain Monte Carlo (MCMC) algorithm, recently proposed in the Bayesian literature, is introduced and applied for a decision-theoretic approach of the optimal weight asset allocation strategy. In particular, the Gibbs sampler proposed by Korobilis (2013) is used in order to estimate the parameters of econometric models for finding the optimal portfolio of an investor. The proposed approach, except for the parameter uncertainty, takes into account the variable-selection uncertainty. More precisely, a computationally efficient algorithm for variable selection is proposed and the approach is compared with the ones in the relevant econometric literature for a managed decision-theoretic portfolio construction.