Watson may have racked up cash on Jeopardy and Deep Blue may have beaten chess champion Garry Kasparov, but if robots are managing your money right now then they’re probably doing one thing: losing it.
These days, most investment funds employ some degree of automated trading. But the sub-category of funds that rely solely on computer software to analyze and bet on markets have had a terrible year. The average such fund is down 3.5%, according to data from Hedge Fund Research (HFR). Hedge funds overall were up 3.9% through August. The S&P 500, by contrast, is up 18.8%.
Investors piled into these quant-driven funds—also known as commodity trading advisors (CTAs)—after the worst of the financial crisis, looking for an edge in grim markets and putting their faith in the physics PhDs and mathematics wizzes hired to work at these firms. The edge for CTAs was supposed to be their ability to make informed decisions at high speed by crunching swaths of data and mining markets for pockets of distortion (paywall), in both bull and bear markets.
Between 2009 and 2012, investors, including pension funds, poured some $130 billion into quant funds. Programmers were only too happy to accommodate them; a record of 187 of these algorithmic funds launched in 2012.
Unfortunately, by that time the profits had already started to vanish. The onslaught of extraordinary money-printing by the world’s central banks changed the assumptions upon which programmers’ strategies were based, which particularly affected the funds’ programming around US Treasurys, which are a primary asset class for those funds. Computers are ill-suited to predict how markets will move as the US Federal Reserve ponders tapering its bond-buying plans, or the European Central Bank ponders ways to keep credit loose. Humans haven’t been particularly good at this either, but statistics suggest that computers have been worse.
Investors pulled $1.33 billion from these kinds of funds in the first half of the year, and those withdrawals are likely just the beginning.