There are so many things we can learn about investing from automated investment services, otherwise known as robo-advisors, that it staggers the mind.
After all, they’re computer-driven platforms imbued with advanced Sharpe, mean variance, and Modern Portfolio Theory that produce “efficient frontier” portfolios, and whose proprietary algorithms rebalance your portfolio based on threshold continuums and your personal dreams and goals.
Oh yeah, they’re complicated.
But there’s one lesson (one you were never taught) you can learn from taking a good, long look at how these robo-advisory services construct and rebalance your personal portfolio.
Because as we’re driven toward a future programmed by algorithms and dominated by Fintech innovations, it’s important to stop and take stock of some of the basic assumptions that will inform that future.
As you’ll see, there’s one fundamental assumption – upon which millions of portfolios have been built – that’s simply dead wrong.
Building a Portfolio Requires Trust
There are multiple theories about how to construct the best portfolio, whether that’s an all-equities portfolio, a cross-asset portfolio, or a personal portfolio built just for you.
Robo-advisors use some of the same theories and math to build you a “personal” portfolio that’s diversified enough that you’re supposed to trust them to manage your money better than you can, better than any human advisor can, better than any other “system” can.
But before you forge that trust, you need to know a few things about how they do what they say they do.
The most important thing being how they construct and diversify your portfolio.
The second thing you have to trust implicitly is your robo-advisor’s ability to rebalance your portfolio, so it doesn’t “drift” away from its “efficient frontier,” but moves your investments around like a master chess player, presumably like IBM’s Watson using your pawn to take someone else’s king.
Theory-wise, robo-advisors start out employing established principles like the “random walk.” That comes from American economist Burton Malkiel’s famous book, “A Random Walk Down Wall Street,” which argues that markets price stocks so efficiently that professionals can’t outperform market indexes. So, passive investing, in an index fund for example, is all you really need to do with your money to play the markets.
Or Wharton School Professor of Finance Jeremy Siegel’s theory from his book “Stocks for the Long Run,” which implies the relative risk of different asset classes depends on the holding period because, he says, stock and bond returns do not follow a random walk. Siegel has shown that stocks actually exhibit “mean-reverting” behavior and bond returns exhibit “mean-averting” behavior.
Beyond single stock price and return theories – because who would ever own a single stock – robo-advisors have to employ portfolio construction theories to diversify clients.
Whether taking a “random walk” or investing in “Stocks for the Long Run,” robo-advisors are mostly all about index investing. While most of the indexed ETF products they put you in are “passive” funds, meaning they’re not managed, some are passive fundamental funds, which are passive in that they aren’t managed, except they are managed in terms of a manager picking which stocks meet the fundamental criteria they think will yield price appreciation or value.
The point here is that while you’re supposed to believe that investing in passive indexes is passive or fundamental to sound money management, by the very nature of how they construct your portfolio and how they move your investments around, robo-advisors are trying to beat the market for you.
To create your portfolio some robo-advisors employ Nobel laureate economist Harry Markowitz’ modern portfolio theory (MPT), with its mean-variance optimization “that takes into account the overall risk of securities (or asset classes), without separating out their systematic and idiosyncratic (unsystematic) components.” In other words, securities relate to one another through a specified pattern of correlation.
Or they might employ William Sharpe’s theories, another Nobel laureate economist, who believes securities correlate with one another through their relationship with the market return, not on a correlation matrix as in the Markowitz model.
Whichever theories robo-advisors use to build a “personal” portfolio, they all rebalance your magic portfolio from time to time.
Generally, robo-advisors employ “threshold-based” rebalancing, which measures asset class or index weightings and whether they’ve “drifted” away from targeted allocations by some amount. An algorithm automatically trades in and out of positions to bring the asset allocation back to target.
Threshold-based rebalancing is one way to go. Another way to reallocate positions might be time-based rebalancing. But most robo-advisors don’t use time-based rebalancing, which can be a problem if your personal investment objectives are time-based and you want to remain in winning positions.
Basically, the idea of rebalancing is about selling overweighted asset classes (because they have appreciated or other asset classes have depreciated) and buying underweighted asset classes to reduce overall portfolio drift.
That’s theory for you. It’s automatic, that’s robo-advisory services for you.
My problem, and I have a lot of issues with how a lot of theories are being applied “automatically,” with threshold-based rebalancing is that, personally, I’ve always made money letting my winners run and cutting my losers loose, as opposed to selling my winners to buy losers. But hey, I’m no robot.
But there’s more.
This is where it gets interesting, so pay attention.
Houston, We Have a Problem…
Whatever the case, all portfolio construction theories – even heavily leaned-on single asset pricing models like the capital asset pricing model (CAPM) – rely on the basic assumption that asset class returns are normally distributed.
That means, statistically, in probability theory, they conform to a “normal” (or Gaussian) distribution, which most of us refer to as a “bell curve.”
There’s your comfort. After all, all of Wall Street bases its financial asset pricing models on a “normal” distribution matrix.
Even the vaunted value at risk (VaR), the mathematical model banks use to determine their exposure to market moves on a daily basis, with close to a 99% confidence level, which regulators rely on, is all based on normal distribution math.
But what if financial asset prices don’t actually conform to a “normal” distribution?
Suprise! They don’t.
That means almost everything we’ve come to believe is a given may not be correct.
I only say may not be correct, because for the most part, financial asset prices, generally on a day-to-day, month-to-month basis, are Gaussian, but in the larger, all-encompassing picture, they are not. They actually conform to a different “wave structure” entirely.
The good news is, the normal distribution models upon which almost all of conventional Wall Street’s thinking and math is based, which are actually subsumed in the larger wave structure that financial prices actually follow, are most of the time how markets roll.
The bad news, which may be increasing based on volatility measures, is that “most of the time” may be interrupted more frequently in the future as larger and larger pools of capital move faster and faster into and out of financial instruments.
Kind of like normal sets of waves rolling along, then – either predictably or surprisingly – being subsumed by a set of tsunamis.
Because distributions aren’t normal, when the tails of Gussian distribution models get fat and blow up, destroying the normal world, revealing the true nature of financial pricing distribution patterns, we wonder what’s flawed.
The CAPM is flawed, MPT is flawed, and VaR is flawed… In effect, the entire basis of robo-advisories is flawed.
That’s your lesson for today.
No, I’m not going to leave it there. Next week, I’ll explain the truth about all this, so don’t go anywhere.