The fundamental objective of conventional Modern Portfolio Theory (MPT) is to allocate your investments appropriately to maximize expected return for a level of risk.  Mean Variance Optimization (MVO) is a quantitative method which allows users to make these allocations offsetting expected return against risk.  The resulting outer convex border of the scatter diagram of potential portfolio combinations is called the efficient portfolio frontier.  The efficient portfolio frontier allows the investor to visualize expected returns per unit of risk for various investment portfolios and to decide where they would like to be positioned along the return/risk spectrum.

This all sounds fairly straight forward and intelligent except for the simple truth; what works in theory frequently does not present well in human reality.  MPT, as with many financial theories, requires a set of assumptions as a basis for application.  In this case, the practitioner requires expected returns and standard deviation of returns (risk) for each portfolio asset candidate as well as a correlation matrix of the relative performance patterns between all combinations of asset candidates.  The most common way of generating these inputs is mining historical data.  However, there are many common sense and quantitative problems with this approach.  Why assume history repeats itself?  Past relationships don’t necessarily repeat.  The theory assumes that returns during different periods are independent of each other and that all return distributions adhere to “normal” statistical construct.  Both assumptions are naive because asset valuations repeatedly show some modicum of reversion after extreme moves and fat-tailed skewedness, though rare, clearly occurs.  The theory holds that correlations between assets are constant.  Clearly as we have witnessed over the last 25 years, the use of leverage combined with open global capital flows have caused episodes in which correlations amongst most, if not all assets, tend towards 1.

We recognize the short comings of MPT, but nonetheless, using MVO with knowledge of theory shortcomings does help us maintain an unemotional discipline  while analyzing index sectors and asset choices as part of our process to create the optimal client portfolio.

1. Gilli, Manfred and Kellezi, Evis, “A Heuristic Approach to Portfolio Optimization” (October 2000 ). Available at View