
Machine-learning and artificial intelligence (AI) techniques are poised to make investment ‘alpha’ easier to find but quicker to disappear, according to a new Robeco paper.
The think-piece authored by Robeco head of next gen research, Mike Chen, says a step-change in technology will increase the ability of quantitative-based investors to identify sources of outperformance in almost real-time – albeit in a trend set to trigger an accelerating competitive arms-race.
“An increase in computing power and algorithms’ ability to make logical inferences, combined with the availability and use of big and multi-modal data, will make detecting alpha signals quicker for all quant practitioners. We’ve seen this with each generation of new technology: the barrier to entry becomes less and less, and the speed of updates is faster and faster,” the report says.
“… This democratization of powerful algorithms for model development will likely lead to faster alpha decay, known as the loss in predictive power of an alpha model over time. Quant research, especially for alpha, therefore, has to be a constant and essential endeavor for quant practitioners.”
As well as speeding up the alpha cycle, the Robeco paper says modern tech also has the potential to dramatically improve risk management, allowing quant investors to immediately tweak portfolios in response to market signals.
Furthermore, the report notes that the advanced tools could automate risk scenario updates “in near-real time by scouring various textual sources (news, social media, company statements, official economic statements, etc.)”.
“Combined to construct knowledge graphs, this information can then be used to enable more rigorous stress testing of portfolios, simulating a wider range of scenarios to understand potential vulnerabilities better.”
But the technology revolution in quant investment might be constrained by a talent shortage in the new areas of expertise or a clash of cultures between old-school practitioners and next-gen tech-heads.
Chen suggests in the paper, too, that AI and other number-crunching technological advances have the potential to replace many quant investment tasks currently carried out by humans over the next 20 years.
However, he says people will still have the power.
“Despite advanced technology potentially transforming the quant investment industry, firms are still composed of people: people with desires, fears, and aspirations,” he says. “Quant firms can only succeed under the right culture, management, and incentivization structure. This part is universal and timeless.”