Last week there was a great piece at ETF.com, “Inside Robo Advisor Asset Allocation.” The author picks through the recommended portfolios from Wealthfront, Betterment, WiseBaynan and other new “roboadvisors.” It was an interesting look and revealed just how different multiple “passive” portfolios can be. I had a few observations that I shared on Twitter:
That piece on robo-advisor allocations shows you that there is no “perfect” portfolio and everyone has a different opinion of “ideal”
— James Osborne, CFP® (@BasonAsset) August 23, 2014
//platform.twitter.com/widgets.js
And WiseBanyan doesn’t think that REITs are an equity asset class. Well they certainly aren’t fixed income. — James Osborne, CFP® (@BasonAsset) August 23, 2014
//platform.twitter.com/widgets.js
Also @wealthfront thinks that dividend stocks are a bond substitute (might have some investors be surprised by that in the next downturn)
— James Osborne, CFP® (@BasonAsset) August 23, 2014
//platform.twitter.com/widgets.js
The latter sparked a conversation with Adam Nash, the CEO of Wealthfront on their use of a dividend-focused ETF alongside a broad market US stock allocation. As a result (and for fun) I spent a good amount of time going through Wealthfront’s “Methodology” page wherein they describe the “science” behind building their investment portfolios. The piece is quite lengthy and thorough. It certainly gives the impression that Wealthfront has this portfolio-building thing nailed. It is filled with lines like these:
“We use Modern Portfolio Theory (MPT) to identify the ideal portfolio for each client.” (Emphasis mine)
“Wealthfront determines the optimal mix of our chosen asset classes by solving the “Efficient Frontier” using Mean-Variance Optimization (MVO) (Markowitz, 1952), the foundation of Modern Portfolio Theory.” (Emphasis mine)
“MVO calculates the best risk-return tradeoff when combining the asset classes in portfolios.” (Emphasis mine)
These statements serve to convince readers that investing properly is a scientific process and Wealthfront’s investment management team has it cornered. The problem with the confidence inherit in these and similar statements is that MVO or any “optimized” portfolio can only exist after the advisor (robo- or otherwise) has provided very sensitive inputs. MVO relies on assumptions of future returns, risk (standard deviation of returns) and correlation between asset classes to determine its output. This is where it gets sticky. You see, no one can predict these three factors for two asset classes, let alone 6 or 8 or 12. So going in, the assumptions are flawed.
And Wealthfront admits it! You see, Wealthfront doesn’t simply take the output suggested by the MVO software and run off to build portfolios. Instead certain overrides are in place because the MVO outputs would most likely provide a very strange-looking portfolio. So minimum and maximum constraints are placed on each asset class. Wealthfront uses a model to constrain inputs that “…largely mitigates the optimizer’s sensitivity problem and enables it to produce diversified and intuitive portfolios.” (Read: forces the MVO software to spit out more “normal” looking portfolios.)
The thing is, I don’t really have a problem with Wealthfront’s process. And this isn’t really about Wealthfront. As I said publicly, I think that the robo-advisors steering investors towards long-term, low cost portfolios of index funds and ETFs is certainly better than the advice the general public gets from their neighborhood broker.
My issue is the illusion of certainly offered by a 5,000 word white paper on a methodology of building investment portfolios. Such a verbose display tells investors “we’ve done our homework” and encourages them to believe that there is a way to know what the “right” future portfolio is going to be, and they are the ones to tell you about it.
Here’s what an honest “methodology” page would really say:
“First, we make a LOT of assumptions about how markets will behave in the future. A good many of these are grounded in historical data, but we like to put our own spin on how that data will or will not be reliable in the future. Next, we put a “Goldilocks” allocation (enough to make a difference, not too much to blow things up) of your portfolio into each one of these asset classes, based on whether you told us you want to take a ton of risk, a fair amount of risk or just some risk (of volatility). We don’t really know which asset class will be best in the future (and thus should get the largest allocation). But we’re going to steer your portfolio in what we think will be best anyhow, and it’s going to look different than the global stock market. We hope we have it right.”
There is always talk in this industry as to whether investing is “art or science.” Those backed by academic research like to hang their hats on science. Many traders, brokers and stock pickers will tell you there is more art to the process. Even academically based strategies such as Wealthfront aren’t 100% “science.” The reality is that, to some extent, we’re all working somewhere in between. I’ll have more in a follow-up piece discussing a possible better way to approaching building investment portfolios than MVO.
As I’ve said before, there is no way to build the “right” long-term portfolio in advance. You (or any advisor) can’t know where the “efficient frontier” lies for the next 20 years. It is much more important that investors build a portfolio they can maintain for 20 years than be lured in to believe that anyone has a methodology to accurately predict the future of markets.
Bason Asset Management 