Efficient Frontier
From Planipedia
The Efficient Frontier is a part of Modern Portfolio Theory. It is a mathematical process, referred to as "mean variance optimization" in which a specific combination of asset classes can be combined that will result in the lowest possible risk for a target level of return, or else the highest possible return for taking on a particular level of risk.
An efficient portfolio is therefore one point on a curve that results if we calculate the efficient portfolio for every level of risk (or return).
The graph above is based on 35 years of data from Canada with 10 different asset classes. Based on this historical data for each asset class we can calculate the historical rate of return and a standard deviation. Each of the 10 year squares represents a different aasset class. You can see the green line which represents the efficient frontier, where the return cannot be increased without increasing the level of risk. The yellow line represents the "inefficient frontier", the worst possible portfolio return for any level of risk.
A major challenge with using the efficient frontier is what is referred to as Sampling Error which is the fact that depending on the timeframe of the data we use to base the calculation on, we can end up with a different answer. It turns out that the efficient frontier is not very effective in "predicting" what is a good portfolio for next year.
Some research - DeMiguel, Victor, Lorenzo Garlappi & Raman Uppal May 31, 2005, How Inefficient is the 1/N Asset-Allocation Strategy - has concluded "The 1/N asset allocation rule typically has a higher out-of-sample Sharpe Ratio, a higher certainty-equivalent value, and a lower turnover than optimal asset allocation policies". In other words if I simply divide up the clients portfolio evenly among different asset classes it will do as well as trying to predict with an efficient frontier.
This has led some advisors to take the approach that since history is obviously not a reliable basis on which to make these calculations that they will make their "best guess" as to the returns or standard deviations of asset classes in the future. The difficulty with this is that it is very "judgemental" and there appears to be little evidence (if any) that it is any more effective an approach than using history.
From a planning perspective, the important lesson is that although the efficient frontier is a "tool", it should not be used blindly and an advisor should always ensure proper levels of diversification.
