What is the primary purpose of resampled mean-variance optimization?

Resampled mean-variance optimization is primarily aimed at improving the diversification of asset allocations. This method involves running a series of simulations on expected returns and covariances to create a range of potential future scenarios. By doing this, it helps to identify asset allocations that are not just optimal under a single set of assumptions, but that also provide robust diversification across various possible outcomes.

The approach acknowledges the uncertainties in input estimates inherent in traditional mean-variance optimization, which can lead to concentrated portfolios that may not perform well in practice. By utilizing resampling, it generates a more comprehensive set of possible outcomes, which allows for the identification of portfolios that can better withstand market fluctuations and reduce risk through diversification.

It is important to note that while resampled mean-variance optimization generates more-diversified portfolios, the process is not primarily focused on creating static asset allocation strategies or minimizing trading costs, nor does it enhance optimization through dynamic factors alone. The core objective is to leverage the statistical properties of asset returns to arrive at allocations that can withstand various future states of the world while balancing risk and return effectively.