Can a computer invest as smartly as Warren Buffett, George Soros?1 min read . Updated: 09 Dec 2016, 11:35 AM IST
Turns out, it can
Turns out, it can
According to a new paper by AQR Capital Management, the program-driven investment firm, that isn’t publicly available, a lot of the gains reaped by the legendary managers over time were, in theory, available to anyone using a handful of buy-and-sell signals known to quantitative analysts.
AQR compared Warren Buffett’s and George Soros’s returns with portfolios automatically tuned to investment styles deemed consistent with their philosophies and found the computers did a fair job of replicating the humans.
“If investors want to be more Buffett-like, they don’t have to pick the exact same stocks that Buffett is picking, they just have to pick the same type of stocks," said Dan Villalon, co-head of AQR’s US portfolio solutions group. “That’s where a lot of the interest behind factors and styles is coming from."
In scanning the market for systems that operate like Buffett, AQR proposed tracking a computer-generated portfolio chosen by the value factor, the quality factor (or stocks with the strongest balance sheets) and the low-volatility factor, meaning those with the smallest price swings. Combined with the market itself, owning stocks with those traits over the past four decades got you within 4 percentage points of Buffett’s results. Buffett and Soros didn’t return messages left with assistants seeking comment.
AQR decided the factors that best capture Soros’s theory on boom-and-bust cycles are trend and momentum. The analysts also included a value factor for currency, a nod to his success in short-selling sterling in the 1992 UK pound crisis. Together, those factors, along with simple US equity market returns, come close to replicating the 20% annual gain in his Quantum Fund from 1985 to 2004, almost triple the US stock market.