Mean-variance optimization (MVO), epitomized by Nobel prize winner prof. Markowitz, has conquered a pivotal place in academic works. Not as much as in asset managers strategies. An asset manager’s approach to investment management is somewhat different from the statistical and mathematical view of securities needed to feed an MVO model.
Furthermore, errors in estimating the parameters lead to poor performances of MVO models when tested on out-of-sample data sets. The rise of quant strategies, combined with the latest robot driven approaches to investment advice, has narrowed the gap.
Any advanced program in an MBA in finance would not be complete without a thorough examination of MVO concepts as well as an application on actual data (the sort in the chart above), though only seldom these lab exercises get carried out with an out-of-sample backtest.
It’s luring to show the rigor of MVO leads to efficient portfolios. Nevertheless, it sometimes isn’t clear that the outcome is the allocation one should have had the year before to have an efficient portfolio. With hindsight optimization is easier.
Leave a Reply