Market volatility, optimal portfolios and naive asset allocations

This paper investigates the impact of a financial turmoil on the performances of traditional, and naive, asset allocation strategies. We compare over a long time span (lasting for the last 60 years) the 1/N portfolio with mean-variance optimal portfolio strategies. Our analyses consider several datasets, and different approaches for the estimation of expected returns, starting from simple historical moments to the use of predictable variables, mean reversion or both. By employing rolling estimation approaches and robust Sharpe ratio testing we determine if during different market volatility states calibrated portfolios perform better than optimally determined allocations.