A Paul Tudor Jones-backed hedge fund that harnesses the stock-picking skills of non-finance quants has nabbed another $100 million of inflows after soaring 20% in last year’s market turmoil.
(Bloomberg) — A Paul Tudor Jones-backed hedge fund that harnesses the stock-picking skills of non-finance quants has nabbed another $100 million of inflows after soaring 20% in last year’s market turmoil.
San Francisco-based Numerai LLC invests based on the wisdom of crowds by collating trading ideas from the likes of freelance data wonks, engineers and scientists, who are then rewarded with its native cryptocurrency. The unconventional business model is riding a slew of Wall Street fads from democratized trading and crypto to artificial intelligence, but so far it’s working.
Since its September 2019 inception, the Numerai One fund, which charges its growing number of institutional backers fees that are close to the hedge-fund industry norm, has returned 49%.
Numerai, also backed by former Renaissance Technologies executive Howard Morgan, now oversees nearly $300 million all-in after fresh allocations from an endowment and US multi-manager, says 35-year old founder Richard Craib. The firm also recently raised $10 million from investors including Union Square Ventures.
“At this point Numerai really has built the next Millennium except it’s an open platform and every strategy is an AI model,” said Craib, not shying away from name-checking his larger, more established peers. “The bigger fact is the consistency of returns versus our peers.”
Numerai’s success comes as Wall Street grapples with a cut-throat battle for quant talent, while ChatGPT’s rise paints an open-for-all vision for AI.
Numerai gives its users abstracted data representing around 2,000 features of every stock, with no one knowing the specific company they’re analyzing. For example, contributors are presented with a number that represents the market return, rather than the actual return itself. As a result, trading ideas end up being based on statistical patterns, rather than driven by a specific economic or market intuition. And more often than not, the amateur finance quants are using machine-learning techniques to comb through the data.
It all makes the fund something of an investing black box, masking the proximate drivers for performance. The fund’s stellar 2022 comes amid a broader quant renaissance as the new inflationary regime proves more favorable for strategies like value and trend following.
Wisdom of Crowds
Anyone can submit their ideas to the crowdsourced fund, as long as they bet on them with Numerai-issued tokens. The firm’s small cohort of in-house quants then amalgamates the stock-market predictions, which are weighted based on those stakes. The firm ensures its overall portfolio isn’t biased in favor of a particular sector, country or factor — a quant term for common stock characteristics like how cheap an equity looks or how fast it’s risen. Those that contributed winning models get more tokens. Losers see their stakes reduced.
“Numerai gets to assimilate the advances in artificial intelligence much faster than other hedge funds,” Craib said. “Some new user will join, have an innovative idea and it immediately gets integrated into our trading strategy.”
There have been few truly crowdsourced hedge funds, and Numerai’s track record is still short. The last well-known example, Quantopian, shut in 2020 as turning a slew of hypothetical strategies into scalable trades in the real world proved difficult. To skeptics, there’s a good reason for this: If even pros regularly struggle to outwit efficient modern markets, how could newbies with less technological resources and experience pull it off?
For Craib, the key to Numerai’s success so far is that participants have skin in the game and are quickly rewarded or punished based on their picks. Most of the consistent outperformers turn out to be advanced data scientists who don’t work in finance in their day jobs.
Winners are paid in a volatile cryptocurrency of the firm’s own making. The Numeraire, or NMR, token has dropped about 34% over the past year, a bit less than Bitcoin, to a total market value of $220 million, according to CoinMarketCap.com.
For now, that decline hasn’t been enough to affect the business model, Craib says.
“If we get to $1 billion in AUM we’re going to have no problem paying our users in any currency they want,” he said. “Right now we like NMR, but in the future it could be any currency or payment.”
(Corrects story published Feb. 28 to amend description of participating data scientists in first deck headline and lede.)
More stories like this are available on bloomberg.com
©2023 Bloomberg L.P.