Goldman's Quant Unit Rebuilt on Lessons Learned in 2007 MeltdownBy
Quants use less leverage, more factors aiming for consistency
Assets climb to $110 billion in smart-beta, big-data offerings
Ten years ago, an abrupt meltdown in quantitative equity funds worldwide shook the burgeoning industry, spurring an exodus of investors. Goldman Sachs Group Inc. was among the worst hit, shedding three-quarters of its $165 billion in quant investments by 2012.
Gary Chropuvka, one of two partners leading the Quantitative Investment Strategies team at Goldman in New York, sweated those hot August days in 2007 that he says he’ll never forget. Rather than losing faith in quant investing, the group began to rebuild the strategies with less leverage and more diversity.
A decade later, the quant unit has clawed itself back to respectability. It manages about $110 billion in big-data mutual funds, smart-beta pools and other products -- all driven by factors such as consumer sentiment from social media. But a humbled Chropuvka now faces more robust competition, with almost every asset manager chasing quant money, betting on similar factors and shaving fees on exchange-traded funds to near-zero.
“We view this as a turnaround,” Chropuvka, 45, said in an interview at the bank’s New York headquarters. “We rolled up our sleeves and rebuilt the quant business based on the lessons we learned.”
During the week of Aug. 6, 2007, as cracks in the subprime mortgage market were just beginning to show, a number of long-short quant managers were suddenly hit with big losses. MIT professor Andrew Lo and a colleague wrote that the meltdown may have been triggered by a rapid unwinding of one or more large quantitative portfolios, possibly the result of margin calls or risk reduction. That led to the “fire sale liquidation of similar portfolios that happened to be quantitatively constructed,” they said in the 2007 paper.
At Goldman, ground zero for the implosion was a hedge fund called Global Equity Opportunities, which lost 23 percent that month. The firm in a report blamed the losses on the use of leverage in the fund and too many copycat managers making the same trades.
“Longer term successful quant managers will have to rely more on unique factors,” Goldman wrote.
Aims for Consistency
Today the QIS unit uses leverage in only some offerings and monitors markets for signs that certain trades are getting too crowded. The group strives for “persistent and consistent returns,” said Chropuvka, and no longer reaches too far for alpha, or benchmark-beating returns.
Smart-beta is the biggest piece of Goldman’s quant business, with about $77 billion in investments that track indexes based on factors such as low valuation and momentum. The unit has several billion dollars in smart-beta ETFs that were among the most successful launches of the past few years, said Eric Balchunas, an ETF analyst for Bloomberg Intelligence.
“Their funds are much cheaper than the competition’s,” said Balchunas.
QIS’s big-data strategies manage about $25 billion in a series of U.S. mutual funds and other products. Quants use machine-learning algorithms to sift through data, including credit-card receipts, to glean clues that might translate into wagers. Chropuvka said they have found 150 different trading signals, increasing the chances his team’s bets won’t mimic those of other quant funds.
Signals sometimes stop working. Tracking changes in analyst recommendations on companies used to be an indicator, but over time it became too widely followed and lost its predicative power. Now Goldman’s proprietary natural-language processing algorithms examine the tone in analyst reports, a more subtle signal that still has value, Chropuvka said.
The big-data mutual funds generally have performed well. The nine U.S. equity Insights funds on average beat 74 percent of rivals over the past five years, according to data compiled by Bloomberg. The unit manages an additional $8 billion of quantitative investments that attempt to duplicate hedge fund strategies at a lower price.
Quants are more worried about overcrowding and a decline in performance than a repeat of the panic of 2007. Rob Arnott, the founder of Research Affiliates and one of pioneers of smart-beta investing, says the rush of money pouring into the category could be its undoing.
“I am urging caution for a simple reason. It is human nature to engage in performance chasing,” Arnott told Bloomberg in June. “People think I am saying smart beta is a bad idea. I am not saying that all, but look before you leap.”
Jonathon Jacobson, founder of Highfields Capital Management, questioned the sustainability of quant performance in a second-quarter letter to investors. “Hundreds of billions of dollars can’t exploit the same inefficiency for long,” the hedge fund manager wrote. “My sense is we are a lot closer to the end than the beginning of these strategies producing excess returns.”
Goldman in a December white paper pointed out that not all smart-beta funds emphasize the same factors, which means the danger of crowding may be not as severe as some fear.
Chropuvka stresses that quant investing doesn’t imply that all decisions are made by machines on autopilot. Managers still set the strategies, make the wagers and monitor risk to try to prevent another quant blowup.
“People play a paramount role,” he said. “We use economic intuition and ask: why does something work and will it work going forward?”
— With assistance by Dani Burger