Algorithms, Robo-Advisors and Investment Automation
...excerpt from Mo Lidsky's latest book, Partners in Preservation
While the rise of robo-advisors has merit for many investors, for affluent investors with sophisticated portfolios investment decisions cannot be replicated with an algorithm, a database, or any piece of technology alone. There will always be those touting the latest and greatest system, claiming that it can solve every investor’s problems with the click of a button. The technology of the future may make me eat my words, but I believe it is unlikely that the subjective art and science of advising affluent investors—dealing with all their specific wants and needs—can be boiled down to a mathematical formula, algorithm, or line of code - any more than raising a child can be boiled down to a robo-parent singing lullabies. There are too many human elements that simply cannot be quantified.
Even if we put the advisory element aside, there is no concrete evidence that any single algorithm can produce consistent outperformance over the long run. In fact, Nobel Prize–winning economist Eugene Fama claims that one of his earliest revelations about the markets came during his last year as an undergraduate student at Tufts University. While working for economics professor Harry Ernst, who also had a stock market forecasting service, Fama was given the responsibility of inventing trading strategies and algorithms to forecast the market. In his efforts, he found numerous formulas that beat the market when back-tested but when put to the test in real life they were practically useless. He found that markets, at least in the short to medium term, are inherently unpredictable.
Fama’s experience puts into question all the newsletter marketers who claim that if by subscribing to their stock picks, investors would have yielded consistent triple-digit returns that exceeded the success of investment legends like Warren Buffett and George Soros by a factor of five. Whether these assertions are based on a rear view mirror perspective vs. actual realized picks can be difficult to determine. It seems subscribers always just miss out on the big return. [It is also worth noting that none of the individuals marketing stock-picking newsletters can be found on the Forbes Billionaires List. Yet, compounding at their alleged rate of return should have certainly gotten them there.]
This reality is illustrated in the writings of Murray Stahl. In Collected Commentaries and Conundrums Regarding Value Investing, Stahl shares a fascinating study where a group of finance gurus were challenged to create a singular model that could have been used (with the benefit of hindsight) to predict the top-performing stocks of the last fifty years. Not one person could produce a model that had the capacity to select the top ten stocks of the last half century. If it is so difficult to create a model that simply reproduced the performance of the known past, how likely is it that another model or algorithm would be effective in making decisions about the future?