Can AI Enhance ETF Portfolios?

Can AI Enhance ETF Portfolios?

Issuers are lining up to launch ETFs using artificial intelligence, joining several already on the market.

JessicaFerringer200x200.jpg
|
Reviewed by: Jessica Ferringer
,
Edited by: Jessica Ferringer

Artificial intelligence (AI) and machine learning are being used to innovate across many different industries. A slew of recent filings hints that the robots could be coming to disrupt the world of ETFs as well.

On Nov. 5, AdvisorShares filed for the AdvisorShares Let Bob AI Powered Momentum ETF (LETB). That same day, WisdomTree filed a fund supplement stating that two of its funds will be seeing a change in name, ticker and investment philosophy in January 2022.

The WisdomTree International Dividend Ex-Financials Fund (DOO) and the WisdomTree U.S. Dividend Ex-Financial Fund (DTN) will be revamped to become actively managed funds that incorporate artificial intelligence and machine learning into the portfolio construction process.

With several ETFs already on the market that incorporate these technologies, is there any evidence that AI and machine learning can add value to the portfolio construction process?

First Active AI ETF

The first active ETF to rely on artificial intelligence is the AI Powered Equity ETF (AIEQ), launched in October 2017. This ETF uses AI to build predictive models on the universe of U.S. equities, identifying 30 to 125 names that have the highest potential for capital appreciation.

As shown in our ETF Comparison Tool, the fund is considerably more expensive than the cap-weighted SPDR S&P 500 ETF Trust (SPY).

 

Pop-up Image

Courtesy of FactSet

(For a larger view, click on the image above)

 

And while the fund has had bouts of over- and underperformance relative to SPY since its launch, performance of the ETFs has been highly correlated.

 

 

While looking at this performance makes it seem like artificial intelligence does not add significant value to the portfolio construction process, it is important to note there are multiple ways AI and machine learning can be applied throughout the investment construction process.

Evolving Technology

Francis Geeseok Oh, managing director at QRAFT AI ETFs, thinks that AI’s role in the portfolio construction process is still in its early stages, and that different iterations of the idea will evolve over time.

“People might think that every artificial-intelligence-based investment strategy may be the same, which is not the case,” he said. “It could be very different [depending on] which machine learning process or technology each individual issuer is applying.”

QRAFT’s AI technology is used within four ETFs, the largest of which is the $30.3 million QRAFT AI-Enhanced U.S. Large Cap Momentum ETF (AMOM).

The AI extracts patterns from analyzing data, and in the case of this ETF, tries to identify stocks that have high residual returns. QRAFT defines this as “its total return after removal of market, size and value risks factored into portfolio construction under conventional portfolio management.”

The fund uses this metric based on the theory that stocks with higher residual returns have the potential to perform better and more consistently over time than conventional momentum stocks. The fund has outperformed the iShares MSCI USA Momentum Factor ETF (MTUM) by a solid margin since its May 2019 launch.

 

 

Using AI To Look Ahead

State Street Global Advisors is another issuer that uses AI within certain ETFs, particularly its Kensho lineup of ETFs. Kensho is the name of S&P Global’s AI and innovation arm.

These ETFs track indices that focus on specific sectors of innovation such as innovative infrastructure, smart transportation and clean energy. The indices underlying these niche ETFs are all subsets of the S&P Kensho New Economies Composite Index, which is tracked by the SPDR S&P Kensho New Economics Composite ETF (KOMP).

The AI used to create these indices relies on a natural language processing algorithm that scans regulatory statements of U.S.-listed stocks. The technology is looking for specific keywords, with the goal of identifying companies that are focusing on these niche technologies.

Matt Bartolini, head of SPDR Americas Research, thinks this technology is useful to identify how companies are evolving their business model.

Discussing the regulatory statements that the AI is analyzing, he said the firm will outline what their forward-looking strategy is.

“If you were to look at something backward-looking, like revenue, revenue has already been realized. It actually is not significantly descriptive of how a firm is evolving along with society,” he said.

KOMP’s AI-driven index has provided outsized returns to the famous active ARK Innovation ETF (ARKK) so far this year, outperforming by 22.6%.

 

 

Time Will Tell

Each of these funds is using AI and machine learning technology in a different way. New filings are likely to have their own twist on the use of these technologies as well.

As time passes, different methods of applying this technology are likely to have more success than others. It is still too early to assess which methods, if any, can provide consistent outperformance over the long run. But as more issuers employ this technology to launch ETFs, this means investors will have ample opportunity to assess whether AI can add value after all.

Contact Jessica Ferringer at [email protected] or follow her on Twitter

Jessica Ferringer, CFA, is a writer and analyst for etf.com. She has 10 years of experience in investment research and due diligence, including helping to manage ETF portfolios. Jessica has a bachelor’s degree in economics from Lafayette College and an MBA from the University of Pittsburgh.