Last year proved to be a strong litmus test for the hedge fund industry, with a slew of strategies able to post solid returns in a market that witnessed an incredible trough to peak recovery between Q2 and Q4.
Last year proved to be a strong litmus test for the hedge fund industry, with a slew of strategies able to post solid returns in a market that witnessed an incredible trough to peak recovery between Q2 and Q4.
But quant funds had a tough time of it, as algorithms struggled to navigate the whipsaw conditions created by the pandemic. This led to significant dispersion of returns, with many household names including Bridgewater Associates, Renaissance Technologies, Winton Capital and AHL Dimension posting negative returns through May 2020. Some quant funds shut down altogether, including Philippe Laffont’s USD350 million quant fund at Coatue Management.
A quick glance at Aurum’s Hedge Fund Performance Review for December 2020 shows that every hedge fund strategy delivered positive returns for the year, apart from quants, which ended 2020 down -5.35 per cent; this is nevertheless a marked improvement on September 2020, when the average quant equity market neutral hedge fund was down -14.02 per cent.
There is no clear and obvious explanation for why quants had a tough time of it in 2020. One might argue that it was the inherent nature of algorithms, which are designed to trade logically and free of human bias, that became their Achilles Heel. The pandemic led to such severe volatility in March and April 2020 that it was hard for algorithms to understand what to make of things. They had no capacity to predict that oil prices would turn negative on 20 April, for example.
Unlike humans, algorithms lacked the ability to think creatively and imagine what ‘could’ happen amidst the chaos of the pandemic.
This proved to be a win for the good old-fashioned discretionary hedge funds who were able to trade the market waves and investor-fuelled perturbations with great alacrity.
This is not to say that quant funds are a busted flush. Far from it. But what 2020 did demonstrate was that never-before-seen exogenous events, on a global scale, can knock even the most sophisticated algorithms off their game and lead to short-term disruption in quant models.
As Adam Taback, chief investment officer of Wells Fargo Private Wealth Management told Bloomberg recently, “Quants rely on data from time periods that have no reflection of today’s environment.”
Michael Weinberg is Managing Director, Head of Hedge Funds & Alternative Alpha at APG, a leading Dutch pension provider. He is also Adjunct Professor of Economics and Finance at Columbia Business School.
In his view, the overall dispersion of returns, not just in quant funds but hedge funds more broadly, showed that alpha can be generated through manager selection. “Last year was a great example of that where you had some quant funds who were up materially while others were very publicly down materially,” Weinberg says.
He attributes the poor performance of quants to the fact that securities that had previously been low beta suddenly became high beta (i.e. consumer product companies), which had an adverse impact on quantitative models. “They hadn’t forecast this would happen, or couldn’t forecast it, and quant managers had a choice to make: either to take their models out of commission or change their model assumptions,” continues Weinberg.
Weinberg spends a lot of time looking at AI-focused investment strategies, also known as Autonomous Learning Investment Strategies (ALIS).
These are strategies that combine big data, data science and machine learning techniques, which, coupled with ever more powerful and cheaper computing power and storage, are pushing the envelope in terms of what is possible in quant trading.
“These strategies continue to evolve,” says Weinberg. “The same trends do not persist. In our view, in the future managers who use alternative data will be able to systematically do what managers have historically done on a discretionary basis, and more efficiently.
“One example of the evolution we are seeing is a manager we met with recently who uses machine learning in litigation finance. Other managers are starting to use AI strategies to analyse and trade mortgage-backed securities.
“We also know one CTA that is using genetic algorithms to trade soft commodities, as well as financial commodities.”
AI-focused hedge funds continue to use different techniques to train their models to better understand the world around us. Weinberg refers to the Book of Why, authored by Judea Pearl and Dana Mackenzie, which explores the new science of cause and effect and how we can think better.
“One manager we know has built a quant AI fund strategy specifically around causality. The strategy is based on a confluence of statistics, probability, calculus, and elements of physics. They believe their AI model will determine a lot of its decisions in a way similar to the way the human brain makes decisions, but with less opacity than using neural networks.”
“Nevertheless, we still firmly believe in ‘person plus machine’. In our experience, those strategies where the coders and the portfolio managers have left the machines to do everything themselves, may have succeeded initially but over time they generally don’t. We would expect the dynamic to continue to be person plus machine.”
APG practices what it preaches when it comes to the power of AI. It is part of the SDI Asset Owner Platform alongside fellow heavyweight investors AustralianSuper, British Columbia Investment Management Corporation and PGGM.
The platform is committed to accelerating the market adoption of Sustainable Development Investments (SDIs). With ESG such a key focus for institutional investors, the platform utilises an AI engine developed by Entis, an APG affiliated entity, to create a new SDI standard for sustainable investing.
“Entis has classified approximately 8,000 global companies based on UN SDGs using machine learning and alternative data in an SDI Asset Owner Platform developed by a group of institutional investors, including APG. Qontigo, the analytics and index provider, distributes the SDI Asset Owner Platform, so that like-minded investors may determine who really is working hard to make the world a better place, and not just involved in greenwashing,” explains Weinberg.
APG is also currently working on an AI research project with Columbia School of Engineering. By leveraging a network of academics and researchers, Weinberg is able to ensure the Dutch pension provider stays closely connected to developments within the AI ecosystem.
Weinberg states that from a high level research perspective the two world leaders, and biggest contributors, are probably the US and China “but from an investment perspective we are seeing less of an opportunity in China. It’s more the US, certain European nations and Israel – that’s where we see most of the talent from an AI perspective”.
“What most excites me is seeing managers deploying new techniques like causality, as I referred to earlier, and the applications to use AI in new markets such as MBS and other structured products. New techniques and data are really beginning to expand the potential for AI strategies in a meaningful way,” Weinberg concludes.
In light of the difficulties that quant funds faced last year, the next few years will be interesting to watch as ever more sophisticated AI funds begin to make their mark.
Who knows…maybe some of the AI strategies that Weinberg refers to will begin to display more of a capacity to think, and imagine like a human, and better respond to extreme events as they unfold in real time.
