Artificial intelligence could change the way ETFs are assembled. Why would an investor opt for traditional index trackers when they can be customized? Newcomers see opportunities, but the establishment remains skeptical.
What trends are emerging within a particular theme, and which companies are benefiting from them? U.S. startup Boosted.ai has developed machine-learning-based algorithms to answer such questions. The model scans thousands of different sources, from news websites to quarterly company updates and documents submitted to regulators.
Speaking to Investment Officer, CEO and co-founder Joshua Pantony said this AI model could have major implications for how ETFs are assembled. For example, an index builder might discover companies he wouldn’t initially associate with a particular theme. “He also gains a much better understanding of why a particular company is linked to that theme,” Pantony said, referring to the comprehensive analysis the model generates.
‘Even greener’
According to Pantony, this creates a new way of building index trackers. ETF providers can easily justify why a particular position belongs in the basket. “This could make sustainable ETFs, for example, even greener than they are now,” he said.
Senior equity strategist Rebecca Chesworth of State Street SPDR ETFs sees that AI is not yet widely used in the composition of ETFs but expects it to be the future. “It’s only a matter of time,” she said.
SPDR uses machine learning only to support information gathering for stock selection for its active funds and theme ETFs. “Thanks to AI, fund managers can act faster and interpret news streams and have better insight into the risk-return profile,” Chesworth said.
According to Pantony, an ETF compiler can now be more efficient. “Previously, it might have taken 40 hours to gather all the necessary information on a particular subject; now it can be done in just 10 minutes,” he noted.
Faster process
Martijn Rozemuller, CEO of VanEck, agrees that AI can speed up labor-intensive processes. “Because of the time savings, you may need to employ fewer people, which brings cost benefits,” he said. However, for an ETF, the savings for investors will be limited, he believes.
“AI can’t time the market,” Rozemuller said. “What I am not convinced of is that AI can also select stocks that consistently beat the market. Humans can’t do that, and neither can AI.”
Chesworth also shares this view: “AI can only recognize some factors that increase the likelihood of outperformance, but there are no guarantees,” she said.
According to Rozemuller, AI can only look at past information, which means it can at best make predictions about the future based on past data.
Gold standard
Pantony believes using AI to create ETF baskets is becoming the gold standard. “At some point, it will no longer be possible for parties to simply present a green ETF without explaining in detail why each investment belongs in the basket,” he said.
Rozemuller added, “We can explain it fine for our products, but sustainable ETFs often spark debates about how green certain companies really are.”
Pantony said AI also makes it easier to customize existing index trackers based on customer requirements. However, Rozemuller is not a big believer in customized ETFs, partly because of regulations.
This article originally was published in Dutch on InvestmentOfficer.nl.
Related articles on Investment Officer Luxembourg:
- Investing with artificial intelligence
- Unlikely startups leapfrog giants in AI goldrush
- Asset managers await a Darwinian consolidation battle