Widely popular with financial advisors + asset managers, our team discusses whether standard deviation (better known as volatility) is a truly effective way to analyze risk in investments, models, and portfolios. To put it simply, volatility just isn’t adequate.
We Begin With the Financial Crisis
During the lead up to the 2008 meltdown, the financial services industry had been creating increasingly complex + opaque products often without understanding their potential risk. Wholesalers of the products, analysts, and even credit agencies did not properly understand the risk of products like CDOs + synthetic derivatives. The result … we know the script, the greatest financial crisis since the Great Depression.
The troubling aspect of the crisis was that many thought these products were safe. In the aftermath, the recommended prescription was to make investing and finance less complex. Yet, that theory continues to fall short. Simple anecdotes do not solve complex problems. By nature, the financial landscape once again is growing more complex. Retail investors and their advisors now have access to complicated managed futures strategies, long/short products, and are increasingly investing into alternatives + hedged strategies.
How many of us understand fully the way that a managed future or long/short mutual fund works? To peel that back, how many financial planners, RIAs, and advisors fully understand the risk of every single ETF and mutual fund you recommend today? The answer for a long time has been using statistics to determine risk through volatility. Is that sufficient?
Past Performance is Not an Indicator of Future Results
Despite a growing level of intricacy in the financial markets, most advisors and risk platforms utilize standard deviation as the primary risk measure to judge suitability of investments. Volatility came into vogue around 65 years ago** **when the concept of using standard statistics to analyze portfolios was introduced by Harry Markowitz. The measure had a couple of advantages.
(Harry himself has argued for expected return to be used in congruence with volatility). Source: http://www.trbimg.com/img-57b8c214/turbine/sdut-harry-markowitz-modern-portfolio-theory-2016jul01
First, intuitively, it made sense. Risky assets were more likely to exhibit volatile swings in value. Second, after the fact, risk measure volatility always added up, **but only after the fact. **
Lets use a visual example to talk about an asset and put it to the test. Here is SPX in 1992-1993 time period:
The chart above illustrates a relatively quiet return series in the S&P 500’s history. During this period the measured volatility of the S&P 500 was around 7.80%. To put that in context, we screened every asset (stocks, ETFs, MFs, Alts, bonds, etc.) in the HiddenLevers database using our Correlation Screener that had volatility currently of 8% or lower + $1B in assets. How many assets did we come up with?
~500 (just north of 1% of the total assets on our platform)
To a casual observer (a client) or someone who is trying to gage the risk of an asset, one might infer the S&P is not a particularly risky investment. Knowing a lot more about the marketplace, though, we know it is not a particularly non-volatile asset. But what about a mutual fund you’re looking at for the first time or a client’s basket of securities. Are you comfortable and familiar with 1000s of products?
No one is and so standard deviation has been an easy way to quickly understand how risky is this asset.
Yet, when we look at SPX over a longer timeframe, the results are starkly different. Not slightly different. Starkly.
Here one can see “after the fact” the S&P 500 exhibits high volatility (~20%) over the past 10 years. While these results should not be surprising, what it does show to us is that standard deviation can be highly influence by timeframe and vastly change. When a statistic can show something in the bottom 1% of risk to something highly volatile, it brings about questions about the methodology of volatility.
Expected Return + Similar Securities
If we have two similar securities with standard deviation do we now have a solid representation of how that security will perform, what it might return, or how it might do under certain economic changes? In fact, we found two securities with very similar standard deviations on HiddenLevers that are vastly different risk profiles, different asset classes, and have significantly different reward profiles.
Security A – BlackRock Global Long/Short Credit Institutional (BGCIX)
Security B – Prudential Short-Term Corporate Bond Q (PSTQX)
Over the past five years, these two securities have had standard deviation of 1.63 + 1.60, respectively. Quite similar. Interestingly, though, BCGIX has produced more than 300+ basis points of return, has a max drawdown over twice as great, costs twice as much, has no yield, and has a much higher Sharpe ratio (risk versus reward). Further, these assets are widely different in allocation. Here’s a snapshot of the Risk Profile Comparison in HiddenLevers:
Further, what does standard deviation tell us beyond simply a snapshot of volatility over a timeframe? Take a security with a standard deviation of 5% and another with 10%. One seemingly is less risky than the other. Yet, what if we also knew the expected return was 3% in the prior security but 14% in the latter. The reality is that we could actually lose more money with the first security than the second even though it has more volatility since the expected return is so low on the first asset.
Is volatility really helpful in understanding risk? Only if it’s used in congruence with many other pieces of analysis. When it’s the driver of risk conversations, it is simply inadequate. Worse, if an advisor makes a recommendation based on standard deviation alone, the client could end up worse off than if they had a more risky asset.
Welcome to the 21st Century
Despite the prevalence of volatility as the primary means to measure risk, better solutions do exist. Portfolio stress testing shows financial professionals the potential downside of investments regardless of their lifespan or asset class. How?
Stress testing looks at each security over a long period of time and measures its sensitivity to a number of economic factors that cause risk – shocks to oil prices, drops in the marketplace, and changes in interest rates or the dollar. It then measures the sensitivity that the security has to those economic levers and as those shocks occur can predict (not tell you exactly) what would be the expected outcome. Additionally, stress testing measures sensitivity, so even if a stock/bond/ETF/MF has not been around for a large drawdown, we can still measure its relationship to various economic levers (as we call them) on a day-to-day basis and have a good indication of how it would react to a major shock.
If the S&P 500 drops 10% and your mutual fund has a volatility of 12%, what happens to your mutual fund? Do you really know? Stress testing can quite accurately tell you.
Additionally, stress testing helps advisors and portfolio managers highlight the differences between securities that may have similar volatility. For example, stress testing illustrates the difference between a high-yield bond fund and an equity fund with similar volatility.
All in all, volatility simply isn’t adequate. As advisors face rising competition from both colleagues and tech, it is time to “stay ahead” of risk, digital platforms, and client fears.
