To gauge the right price for a rare vintage Bordeaux, you could travel around the French countryside asking grape growers to recall weather from decades ago. The most valuable vintages all follow a similar pattern. Grapes that were grown during a hot and dry summer have a higher sugar concentration, rather than wetter conditions which dilute the flavor.
But one clever economist thought there was a better way. In the 1980s, Princeton labor economist and wine lover Orley Ashenfelter created a formula to predict the future prices of wine based on data, not human intuition. His formula was simple:
∆ price = -12.15 + (β1 * Winter rainfall) + (β2 * Average summer temperature) + (β3 * Harvest rainfall) + (β4 * Age of Vintage)
Why? Ashenfelter’s simple formula meant a merchant could determine the quality of a wine long before it was drinkable, thus turning the world of wine ratings and prices on its head. Surely you need expert intuition, with years of experience drinking lots of wine to be able to assess wine quality? Not at all, it turned out.
Algorithms answer big questions
An algorithm, a process for calculating a range of inputs to come up with an output, can make hard-to-predict events–whether future wine prices or the stock market–more predictable. This is because algorithms don’t get distracted by irrelevant information, short-term fads, or social factors. In fact, algorithms today are able to beat humans in many specific areas, including chess, Jeopardy, baseball, and medical diagnoses to name a few.
At Betterment, we use algorithms to optimize our portfolio—that is the basis of our advice. No intuition, no insider connections, no hunches. Just math.
Shifting a whole industry, one formula at a time
In his recent book Thinking, Fast and Slow, Daniel Kahneman, who won the Nobel Prize for economics in 2002, writes that in nearly all fields where predictability is difficult, everything from sports to school admissions and market pricing, experts are often inferior to algorithms. “Whenever we can replace human judgement by a formula, we should at least consider it,” Kahneman writes.
“Whenever we can replace human judgement by a formula, we should at least consider it,” Kahneman writes.
Yet, much like our old-fashioned wine experts, the traditional financial industry has been slow on the uptake to use algorithms in individual financial advising, even when big banks are using them at the corporate level. Why? In part, many investment services use legacy systems that are expensive and complex to improve. It is easier for big institutions to maintain the status quo.
Then there is the question of the value of certain fund managers and advisors. The transition to algorithmic financial advice is likely to alter the role of those who have traditionally made a living making decisions based on historical performance or who earned commissions based on selling certain products. Many financial advisors do provide far more value than an algorithm could, but for those who don’t, this is a game-changer.
Lastly, there is an innate cognitive bias which leads us to prefer the advice of people over that of software—even when it’s to our detriment.
Investing should not rely on imperfect memory
Human memory is biased, and we are inconsistent at summarizing complex information, Kahneman says. The economist demonstrates again and again in his book that human judgement is not a substitute for a precise formula. For example, convicts often get different treatment for the same crimes because of variations in human judgement.
When it comes to something as significant as your life savings, the goal should be to minimize any margin for error or judgement calls. You should be able to accurately hit the mark you are aiming for with your investments (provided you do your part to save). That’s what Betterment’s automated investing does for you.
Indeed, Betterment’s algorithms do things with a level or precision that human investment advisors or DIY investors cannot match, including trading in fractional shares and automatically rebalancing your investments with dividends in a tax-efficient manner. Not only can algorithms handle what humans once did: they can do it better.
Soon after they were harvested— and long before they could be tasted—Ashenfelter predicted that the 1989 grapes would be an excellent vintage. The experts grumbled at the time, but Ashenfelter’s formula proved right—that vintage was deemed excellent by those same tasters a decade later.
To be sure, the wine world has now come to accept technology, including pricing algorithms and metrics, as part of the industry. The same is true for investing: based on what we know to be true, we believe our algorithms can manage your investments better than manual investment services.