Ben Carlson’s concise evaluation of mean-variance optimization:
One of the students asked for my thoughts on the efficient frontier and mean-variance optimization. I told them that the general idea behind these theories has been very helpful to the portfolio management industry in a number of ways. Diversification and the idea that adding together investments that behave differently in a portfolio is an important concept.
But you can’t take these types of models literally. Correlations and market relationships are constantly changing. Nothing is stable and the past isn’t a perfect window into what’s going to happen in the future. The efficient frontier shows you the best risk-adjusted returns from a historical data set. It can’t tell you what the perfect asset allocation will be in the future.
Models and textbook theories can play a role in building your knowledge base, but they never tell the whole story. Many people make the mistake of taking them at face value without thinking through the real world implications. No model is perfect, so the majority of the time what really matters is the interpretation by the end user.