Article tiré du magazine The Economist.
Amateur coders write algorithms to compete for funds.
“Quant” hedge funds have long been seen as the nerdy vanguard of finance. Firms such as Renaissance Technologies, Two Sigma and Man AHL, each of which manages tens of billions of dollars, hire talented mathematicians and physicists to sit in their airy offices and develop trading algorithms. But what if such talent could be harnessed without the hassle of an expensive and time-consuming recruitment process? That is the proposition Quantopian, a hedge fund and online crowd-sourcing platform founded in 2011, is testing. Anyone can learn to build trading algorithms on its platform. The most successful are then picked to manage money. Last month the firm announced it had made its first allocations of funds to 15 algorithms it had selected.
Quantopian would appear to have one striking advantage over its competitors: sheer weight of numbers. The difficulty of hiring and a desire for secrecy limit even big quant funds to a full-time research staff in the low hundreds (Man AHL, for instance, has 120). Quantopian boasts 120,000 members on its platform.
These are amateurs, however, not full-time employees. John Fawcett, Quantopian’s CEO, says many sign up to learn how to apply algorithms to trading; they usually already have experience in coding and modelling in domains outside finance. Few will have their algorithms selected, an honour that comes with a licensing fee of 10% of net profits . The rest can at least use their algorithms to trade their own money.
Mr Fawcett plans both to allocate funds to more algorithms, and to increase allocations to those already picked. There is no dearth of capital. Steve Cohen, a big-name investor who survived an insider-trading scandal at his previous hedge fund, provided some of Quantopian’s venture-capital funding and has pledged up to $250m to promising algorithms on the platform. The firm intends to launch a fund open to other investors this year.
Quantopian-like models have the potential to bring the gig economy to high finance. Most people on its platform hold full-time jobs or are students, earning some income on the side. At least one quant hedge fund has already bet on the trend. WorldQuant’s WebSim platform, like Quantopian’s, offers access to financial data and a place to test out building algorithms. The best performers on WebSim can become paid part-time research consultants, of whom there are now close to 500, nearly as many as WorldQuant’s full-time staff.
It is still early to judge Quantopian’s allocations (ranging from $100,000 to $3m per algorithm) by their financial return. As a pioneer, it has no obvious comparators. Some algorithms at Quantiacs, a competitor with only around 6,000 members on its platform, have generated up to 40% returns in the past year, but that is with small allocations of capital (Quantiacs has yet to manage outside assets). So the real test for the crowd-sourcers lies ahead: will a deeper talent pool mean better performance, even when serious money is involved?