Managing Portfolio Risk*
Risk is one of the least understood concepts when it comes to investing, so it is time we insert some discussions into the mix of portfolio construction. I am going to quote Mark Hebner and link to his outstanding web site frequently as he does an excellent job of describing and defining risk. Let us first begin with a definition of Standard Deviation (SD) as it is frequently used to measure portfolio risk.
My mathematical dictionary defines SD as "the square root of the arithmetic mean of the squares of the deviations from the mean." This definition permits one to write an equation to calculate the SD, but it is not too helpful for the average investor. Think of SD as a statistical measure of the historical (generally three years of data) volatility of an investment. Hebner writes, "More specifically, it is a measure of the extent to which numbers are spread around their average. It also quantifies the uncertainty in a random variable, such as historical stock market returns."
A "six sigma event" is a very rare event. Even a three or four sigma event is unusual. Just what does that expression mean? One standard deviation away from the average accounts for approximately 68% of the annual returns in the time period. Suppose I gave a test to 100 students and the average score was 78% out of a perfect score of 100%, and the standard deviation was 12%. That means that 68% of all the scores will lie between 78% -12% or 66% on the low side and 78% + 12% or 90% on the high side. Sixty-eight of the students will have scores between 66% and 90%. If we go to the second standard deviation, 95% of all scores will lie within two standard deviations and 99% will lie within three standard deviations. One can begin to see how rare a "seven sigma" or 7 SD really is. After a three standard deviation we may as well stop calculating the percentage as we are into the granules of that last percentage point.
Let's put this into practical terms with the "Risk Capacity 100 – Bright Red" portfolio, found on page 245 of Hebner's book. If one looks at the 50-year period from 1957 through 2006, this portfolio (high risk) returned an annualized 14.22% with a standard deviation of 15.37%. [The value will change if looking at the on-line data.] I consider anything over 12% or 13% to be getting a tad high in risk. Above 15% is cause for concern and one should consider ways to tamp down the portfolio risk, volatility or uncertainty.
Let's move down to the "Risk Capacity 90 – Gold" portfolio. Over the same 50-year period, 1957 through 2006, the annualized return is a little lower at 13.5% and the standard deviation is also a little lower at 14.15%. [Again, this value is in flux when looking at the on-line data.].
When setting up asset allocations, one needs to decide if the extra return merits the additional risk. The AA-Mosaic and Curie Portfolios closely approximates the "Risk Capacity 90 – Gold" portfolio. We now have software available that gives us guidance of the projected annual risk we are taking on when we build a portfolio using specific ETFs and/or stocks. While most investors focus on return, here at ITA Wealth Management we consider risk to be equally important and this is why we pay a lot of attention to the Information Ratio, Sortino, and Retirement Ratios.
The Sortino and Retirement Ratios are now built into the TLH spreadsheet. These ratios have the advantage of incorporating a semi-variance calculation rather than a mean-variance calculation for risk determination.
To manage risk requires knowing what it is and then what ETFs to use to bring down the projected risk. Going to cash is one method, but that requires market timing – not easy to pull off. We are employing a Tactical Asset Allocation timing model called – ITA Risk Reduction model. Using more bonds is another tack one might take. In certain situations we can use short ETFs such as SH or SDS. That also requires some timing expertise. Our best effort is to initially build portfolios that have projected risk values as low as possible while generating projected returns in excess of our benchmark. We use QPP analysis as an aid in this process.
* This blog entry was revised and brought forward from when it was written several years ago.