I am one of the many advisers who apply modern portfolio theory in our investment strategy and use carefully crafted portfolio allocation models. For each risk level, a model will be constructed of specific percentages of asset classes based on so-called “efficient frontier” calculations designed to produce the greatest return.
That's all well and good, but how do we determine the efficient frontier or, stated more simply, how do we calculate our allocation models?
Using an optimizer does not produce black-and-white answers. If left unconstrained, or without guidelines, the software often recommends portfolios consisting of high percentages of a limited number of asset classes.
For example, one year, my optimizer recommended a portfolio containing all small-cap-value equities and emerging-markets bonds. Another time, the optimized portfolio was heavily weighted toward real estate. I certainly couldn't use models like these for my clients.
So we must constrain and define parameters — limiting real estate, small-cap-value and emerging-markets allocations (among others) — to create models that can be somewhat consistent over time and be palatable to our clients.
This demonstrates that calculating optimal portfolio allocations requires a combination of art and science, or truly, opinion and software. At my firm, we apply this combination approach to asset allocation. Although we never materially change our models (since we are not market timers or strategic allocators), we still consider the relative changes in correlations each year to make minor adjustments.
I've tried various optimizer programs. Over the years, I've used Morningstar EnCorr, Zephyr, Advisory World ICE and others. There are broad ranges of prices and functionality. It's difficult to find the “right” solution. I'd like to hear from you. How do you determine portfolio allocations and what software do you use?
Sheryl Rowling is chief executive of Total Rebalance Expert and principal at Rowling & Associates. She considers herself a non-techie user of technology.