A painful March for traditional quant players makes the 2.6% return, without leverage, put up last month by Castle Ridge Asset Management — a small Toronto-based hedge-fund manager that is run by an artificial intelligence system known as Wallace — that much more impressive.
Big-name managers like Bridgewater, Schonfeld Strategic Advisors, Point72's Cubist, and Renaissance Technologies were slammed in March, and some quants that spoke to Business Insider said it felt worse than the "Quant Quake" of 2007.
Castle Ridge was able to avoid the pain, as its "self-evolving" system that trades large-cap equities even made money on two of three big sell-offs last month, according to a letter the firm sent to investors.
In an interview with Business Insider, Adrian de Valois-Franklin, the firm's CEO and co-founder, said the goal was to create private equity-like returns with hedge-fund liquidity.
The strategy, which currently runs roughly $30 million, made 17.9% and 15.7% in 2019 and 2018, respectively, and is up 3.6% through March for 2020. That is compared to the average losses of 9.4% for the first quarter, according to Hedge Fund Research, while the overall equity market shed roughly a fifth of its value during the same stretch.
The firm subscribes to the "nowcasting" strategy espoused by AQR's former machine-learning head Marcos Lopez de Prado and others, which states that quants should be focused on reacting quickly to new information instead of building models and trying to forecast the future. The goal of AI funds is not to become a better version of quant funds but instead a better version of human, discretionary investors.
"You and I, as fundamental investors, can maybe juggle seven different variables at the same time," de Valois-Franklin said. The firm's Wallace system, he said, juggles thousands. Forecasts are "fundamentally static, despite markets changing all the time."
"It'd be like teaching a computer how to play chess but then the rules of chess change."
A former member of the Canadian Public Pension Investment Board, KKR tech investor, and Goldman Sachs banker, de Valois-Franklin started Castle Ridge with Alex Bogdan, the firm's chief science officer, who has a PhD in artificial intelligence, and they began trading with live capital in 2013. They have since received backing from Canaccord's Genuity Group to create separately managed accounts for wealthy investors.
The expectation from the firm's founders is for machine-learning and artificial intelligence to grow across finance and break away from how it is currently being used.
"The view of machine-learning today is very narrow," Bogdan said.
"It's time to differentiate the static models from the super adaptive."
Many firms use a simplified version of artificial intelligence for risk management, de Valois-Franklin said, but not for idea generation or trading. Naturally, the firm expects that to change — and for those lagging behind to be left behind.
"Those old techniques have no place," de Valois-Franklin said about the traditional quant players.
