Financial advisors across the country have been swarming to software packages designed to assess the risk and return potential of an investment portfolio. Through the magic and science of increased personal computing power and statistics, any individual financial advisor now has access to potent tools once reserved for the largest of institutions.
This type of software comes in many shapes and sizes, with a variety of underlying assumptions and equations driving the output. Some are truly professional level tools founded on spectacular data aggregation and rather complex statistical analysis. Others could be relatively easily recreated in Microsoft Excel by a reasonably competent analyst, given enough time and patience.
For the most part, though, the versions of the software that have caught on are those that – regardless of their theoretical underpinning – allow a financial advisor to show a prospective client the degree of loss they might experience under a variety of adverse economic conditions…and then show them how much better off they might be if the portfolio were managed by the presenter instead. Not surprisingly, these tools first became popular while investors were still quite sensitive to the last financial crisis.
In a vacuum, some of the best versions of this software have been a tremendous boon to independent financial advisors, investment managers, and analysts. Information that might have cost tens of thousands of dollars per year a decade ago, can now be had for a couple hundred dollars a month. With an understanding of intrinsic limitations, the knowledge that can be gleaned from these tools can be leveraged to genuinely improve investment research and risk management processes, streamline security selection, and explore dynamic economic relationships.
Unfortunately, in the world of personal finance, understanding the limits of statistical analysis doesn’t sell subscriptions. What does sell subscriptions are tools that make a financial advisor money, save a financial advisor money, and/or save a financial advisor time, so they have more of it to make money.
Since arriving on the scene several years ago, the justifiable purpose of these software packages lost some its allure. Due to a need to beat out competing software, and a desire to keep subscribers on the hook, the products that have survived have continued to stretch the application of the underlying math and science in the pursuit of a compelling value proposition. What was once little more than really handy information that was fairly innocuous regardless of the user, is quickly turning into something quite frightening.
Financial advisors, regardless of their technical skill level or background, are now utilizing some of these software packages as their beginning-to-end investment management process. Some programs now enable financial advisors to set parameters for optimization (maximum periodic drawdown, average standard deviation, yield, average return, etc.) and the software will output a recommended portfolio from a universe of securities, either uploaded by the advisor or built into the software.
Even this, in and of itself, is not an awful prospect. In fact, the idea of optimizing an investment portfolio using similar means provides the foundation for popular traditional economic models. It is, however, the apparent lack of understanding by some of the software’s users about how these optimizations are derived that might cause some issues.
In certain professional circles, it is not uncommon to hear financial advisors speak of these tools as if they possess predictive power. Perhaps more startling is that some advisors seem to have unabashedly outsourced any semblance of investment due diligence to their trusted software, accepting the output as if it were the only reasonable combination of assets, given the optimization parameters.
It seems that the issue is that some of those relying so heavily on these tools do not understand the limits of what is under the hood. Yes, it is impressive. Yes, it does allow users to peak into the relationships and correlations of variables. Yes, it is tempting to project those relationships into the future as fact (and some versions of the software border on implying that this is reasonable to do). Yes, the engine that powers this modern miracle is historical data…
Oh, yeah, everyone knows that, right? These programmatic panaceas are nothing more than an analysis of historical data. No, sadly, they are not a collection of crystal balls. In the right context, though, this information is extremely valuable. The problem is that this information seems to be increasingly taken out of context.
These software packages do what they are programmed to do, and they do it well. The problem is that they can only work from the data available, and they can only look backwards.
One example of the type of errors that are a typical byproduct of the limitations of this type of software include oil related master limited partnerships (MLPs) being stuffed into portfolios optimized for “conservative” income investors throughout 2015, only to meltdown in spectacular fashion within a year to the great surprise of financial advisors and their clients.
From the software’s perspective, the risk/return characteristics of MLPs over the few years preceding their inclusion into recommended portfolios made them a shoo-in. They produced consistent and attractive returns and little downside variance. A similar phenomenon is presently occurring with high yield bonds seeping into progressively lower risk tolerance portfolio recommendations.
While a knowledgeable human investor could identify assets based on other characteristics that might include or exclude them from consideration, or, at least, modify a recommended asset weight, the software only knows the numbers.
“Past performance does not necessarily predict future results” is a disclaimer mantra in professional investing. What many financial advisors that are over reliant on these tools might not realize is that they are fooling themselves and their clients into believing that past performance does indeed predict future results. The cause for concern is that these same advisors might not understand the evolving nature of how their software develops recommendations, or the truly terrible consequences that can result from malappropriated investment advice.
Coupling the information derived from these tools with a diligent and knowledgeable professional is hard to criticize. However, financial advisors completely sacrificing their own due diligence to rote – and recency biased – statistical processes might be setting themselves and their clients up for failure.