Optimal Portfolio Allocation Under Volatility-Return Regime Switching

Abstract

We study the optimal portfolio allocation when returns, covariances and volatility switch between normal and turbulent regimes. We isolated all the high-volatility events in recent historic return series and produced covariance-matrices corresponding to normal and bear-market regimes. Using these covariance-matrices, along with analysis of the probability of occurrents from these events, led us to produce Monte Carlo simulations with time series of returns used to optimize allocations using investors CRRA-levels, maximizing their utility. Our findings support research showing substantially different allocation when the bear market regimes are taken into account, with heavier weights in lower risk assets, compared to the full-sample covariance-matrix simulation producing much heavier weights in high-risk assets. Our results show that as returns are decreasing with the more conservative allocation, the risk is reduced substantially. Our results also indicate heavy support for the theory of time diversification.