Joint tails impact in stochastic volatility portfolio selection models (Q827150)

The authors investigate the construction of a nonparametric Markov chain process that allows to model the evolution of the cumulative wealth and its empirical volatility over time under thorough a non parametric Markov bivariate process. They use a tree structure to represent the investment evolution under the Markovian hypothesis, and apply a nonparametric Markov chain process to approximate the portfolio return distribution and its volatility under a real world probability. Moreover, they exploit the recombining effect significantly reducing the problem complexity. This tree structure allows an appropriate analysis of the portfolio return and its empirical volatility joints tails. In this framework, the impact of the tails on the optimal portfolio choices is analyzed. Summarising, the main contribution of this work is twofold. Firstly, a non parametric Markov stochastic volatility process is introduced and applied to portfolio selection problems. Secondly, the authors empirically examine the impact of the joints tails of the stochastic volatility model on optimal portfolio choices.