A Crowdsourced Quant Fund Fizzles in Era of Democratized Trading
(Bloomberg Businessweek) -- In an Italian town about 120 miles northeast of Rome, Emiliano Fraticelli spends half his day teaching computer science at a local high school and the other half pursuing a dream he once considered lost to him forever: quantitative trading. He creates computer algorithms that scour market data and make trades based on those patterns.
That’s the kind of thing typically done by professionals working for hedge funds, with sophisticated computers and data feeds at their disposal. Fraticelli, 34, who still lives in his hometown in Teramo, Abruzzo, nestled between mountain ranges and the Adriatic Sea, decided he couldn’t leave his elderly parents to pursue an investing career. “I wanted to have some exposure to this quant world, but I wanted it to be remote,” he says. Then he discovered Quantopian, a Boston-based startup with a free online platform for developing and testing algorithmic strategies.
Quantopian, backed by hedge fund billionaire Steve Cohen and venture capital firm Andreessen Horowitz, was trying to crowdsource great investing ideas. (Bloomberg LP, which owns Bloomberg Businessweek, is an investor in Andreessen Horowitz.) It gave Fraticelli and 300,000 other users a way to try their hand at computerized trading. Those whose programs survived a meticulous screening could have them included in a hedge fund Quantopian ran, and get a cut of their strategies’ profits. The website also hosted contests that gave cash to the top performers. Fraticelli says he won a few thousand dollars.
But now he and his fellow Quantopian users are hunting for an alternative to keep their ambitions alive. In late October, the company announced it was shutting down. A few weeks later, Quantopian Chief Executive Officer John Fawcett announced that he, his co-founder, and other employees were going to work at the retail brokerage Robinhood Markets Inc.
To some pros, the end of Quantopian was inevitable. Could amateurs really figure out anything they couldn’t? Even high-priced hedge fund managers are struggling to outwit the market these days. “If you needed surgery done in a hospital next week, would you let someone who’s just read books on medicine do it?” asks Mathew Burkitt, a former quant at BlueCrest Capital Management who shut his own hedge fund four years ago.
Quantopian’s bet was that this kind of elitism might give it a competitive edge. By offering everyone on the internet free access to data, tutorials, and tools, it sought to beat the army of Ivy League Ph.D.s by picking the best quant strategies from the world’s untapped geniuses. It was the wisdom of the crowds, applied to the nerdiest corner of Wall Street—radical, sure, but a logical extension of a burgeoning gig economy and a tech revolution that was opening up access to ever-deeper market data.
The startup, which was launched in 2011, also tried to make money by selling an enterprise version of its online platform to financial firms. But that never really took off, and it was mainly banking on its hedge fund to succeed, according to people familiar with the matter who spoke on the condition of anonymity. The firm had about $50 million in venture funding, according to Crunchbase. Cohen himself committed as much as $250 million to be managed by the firm.
The fund stopped trading at the start of 2020. In an interview with the Boston Business Journal, Fawcett said the fund had underperformed. He didn’t respond to messages seeking comment. A spokesperson for Robinhood says he and the team from Quantopian will help enhance the information resources available to its customers.
There’s an irony to Quantopian’s people moving to Robinhood. That company’s commission-free trading app has become a phenomenon that’s pulled young retail investors into a booming bull market. One take on Quantopian’s failure is that it’s a lesson in humility for novices hoping to go toe-to-toe with professional traders.
Another is that running a successful hedge fund is much more than amassing trading ideas. Quants perform sophisticated analysis on huge amounts of data to find potentially lasting patterns, and then have to turn those insights into workable trading strategies. Quantopian gave users the tools to hunt for patterns—like the relationship between a stock’s social media mentions and its performance. The next step was putting them together in a profitable way, and that proved difficult.
The platform allowed its users to try almost any strategy. This led to more than 12 million so-called backtests on the platform, in which hypothetical strategies were run against historical data to see if they’d work. But the fund was limited to using a subset of strategies that fit with its particular investing style. Also, many of the users’ strategies were not scalable, meaning that not much money could be invested in them, according to a person familiar with the matter.
Karl Rogers, the founder of hedge fund consulting firm ACE Capital Investments, learned quant trading himself on Quantopian. But he says there just wasn’t enough skill out there for the fund to take advantage of. They were “getting people who just want to learn trading signals or people who don’t do this on a full-time basis and they’re competing with people who do this on a full-time basis,” he says.
“To find positive returns that beat the market and to have to find it in a very specific way makes the problem even harder,” says Jared Broad, founder of rival platform QuantConnect, which makes money by selling its product to financial institutions and running a marketplace where users can offer their algo strategies to anyone who wants to buy them. Crowdsourcing also lives on at other platforms. Numerai, which rewards its users with its own cryptocurrency token, probably comes the closest to Quantopian’s vision.
Professional investors can’t gloat too much, because hedge funds in general are hurting. They’ve lagged the S&P 500 by 62 percentage points over five years, Hedge Fund Research data show. And quant investing in general is full of pitfalls. One is that backtesting can unearth a lot of random signals that don’t have anything to do with why a stock went up or down—they might appear to have predicted moves in the past but won’t in the future. As the availability of data makes it easier to try out hypothetical strategies, investors tend to pick up more of this noise. In a 2016 paper, four Quantopian employees found that the more backtesting a quant did, the bigger the gap between the reported results and the real-world returns.
All quant investors are racing against a market in which the best strategies quickly become open secrets. The democratization of technology and data makes it easier for people to get started in quant investing, says David Khabie-Zeitoune, chief executive officer at GSA Capital, a $4 billion quant hedge fund. “But against that you have a stronger force, which is that there are so many people trying to do this,” he says. “It has never been as ferociously competitive in quant markets.”
James Veitch, a 20-year-old computer science student, hopes to one day join the competition, and he will have Quantopian to thank. The intern at hedge fund Balyasny Asset Management says he first learned to code by editing other people’s work on Quantopian and ran more than 30,000 backtests over four years. Already, he has mastered the ageless rule of hedge funds: Asked about some of his successful trading ideas, he declined to elaborate. Amateurs can crowdsource. Pros keep it to themselves.
©2020 Bloomberg L.P.