The equally weighted portfolio still remains a challenging benchmark

This research replicates the paper "Optimal Versus Naive Diversification: How Inefficient is the 1/N Portfolio Strategy?", DeMiguel et al. (2009b). Similar to the referring paper, working in the mean–variance context, we compare the out-of-sample performance of the same investment strategies on the basis of standard metrics (Sharpe ratio, certainty equivalent and turnover). We consider proportional transaction costs and estimation rolling windows of limited length. Our study updates the original paper for many interesting aspects. First, to exclude that the empirical evidence of DeMiguel et al. (2009b), whose data stopped in 2004, could depend on very specific market behavior, we use an updated version of the original databases that contains the returns of the last 20 years. Recent data are characterized by a few severe systemic events, the 2008 global financial crisis and the shock related to the pandemic, and a generally higher level of price volatility than the previous periods. In our opinion, this variation in the market's conditions makes the replication very interesting. Second, we introduce the Equally Risk Contribution (ERC) portfolio within the allocation strategies under comparison. This allocation rule is strictly related to the mean–variance approach when the variance is used as the referring risk measure and it constitutes a very interesting alternative investment benchmark. Moreover, using real data, we study whether a variation of the holding period or the length of the estimation window can modify the performance of all the strategies under comparison. Our findings confirm the results of DeMiguel et al. (2009b), i.e. that the equally weighted portfolio still remains a challenging benchmark to beat. Nevertheless, we find a few significant differences: the number of strategies that outperform naive diversification is larger due to the increased market volatility; limiting the impact of transaction costs by investing in a portfolio with a stable allocation as the ERC, or modifying the lengths of the estimation window and the holding period, is not sufficient to beat naive diversification systematically.