'So far, AI isn't poised to steal portfolio managers' jobs'
The hunt is on for the magic artificial-intelligence model that will enable investors to beat the stock market. For instance two researchers at Hong Kong University recently published a research paper claiming - in theory, anyway - they may have found one. (They will certainly not be the last.)
When Wall Street can no longer peddle $1.75 trillion space stocks - complete with data centers in space and "asteroid mining" (see Page 10) - their salespeople will come knocking with a new way that AI can supposedly make you rich: by picking stocks.
But, when they do, treat their promises with a healthy degree of skepticism. Beating the market in theory is a lot easier than beating it in practice, and if one person comes up with an artificial-intelligence model that genuinely gives an investor an edge, there is nothing to stop others copying it. Everyone cannot beat the market.
A case in point: Wall Street fund companies have launched over a dozen ETFs that aimed to use AI to pick securities, such as stocks, and beat the market. The results have been disappointing.
"So far, AI isn't poised to steal portfolio managers' jobs," says Bryan Armour, a fund analyst at Morningstar. Of the 16 AI-powered ETFs that have launched since 2017, he says, "most significantly underperformed" the basic Vanguard Total (U.S.) Stock Market ETF VTI, and "half have closed."
"The average AI-powered ETF underperformed VTI by 5% annualized over their lifetime, with four trailing by more than 10% each year on average," he adds.
Among the problems, he says, is that AI-powered ETFs often fell into the "classic pitfalls" of active management: "overtrading and high costs." Many of these funds churned their portfolios too much, responding to short-term signals instead of longer-term trends, he adds.
Innovative fund giant WisdomTree, for example, launched two in January 2022: the AI Enhanced Value Fund AIVL, which uses AI to help pick U.S. value stocks, and an international equivalent, the AI International Enhanced Value Fund AIVI, which uses AI to pick non-U.S. value stocks.
But both funds so far have markedly underperformed simple low-cost index funds that invest blindly in value stocks based on their market capitalization. Since its launch, the AIVL fund has generated total returns of 44%. The Vanguard Value ETF VTV over the same period: 67%.
The international version, AIVI, has generated total returns of 67%, but over the same period the iShares International Value ETF index fund IVLU has earned you 90%.
Among the first AI-powered ETFs available to U.S. investors, Amplify's AI-Powered Equity ETF AIEQ, was launched back in October 2017. Since that time it has generated total returns of 96%. That seems decent, until you realize that over the same period the Vanguard Total U.S. Stock Market Index fund VTI has generated total returns of nearly 172%, or almost twice as much.
I'm not just trying to take cheap shots at either company. Innovators deserve credit for trying. The only guaranteed way to avoid making mistakes is not to do anything. It's not as if other fund companies have done better. Most of them haven't done anything in this area at all. (Of course, that may be cold comfort for those who owned the funds.)
Both Wisdom Tree and Amplify were willing to talk to me about their experiences so far running these funds. And they had a number of things in common.
Most importantly: We are in the very early stages of artificial intelligence, and so it is far too soon to draw long-term conclusions.
"In the early years there were a lot of learnings," says Chris Natividad, chief investment officer at QuantumStreet AI, the technology firm whose artificial-intelligence model powers the Amplify fund. "AIEQ ... is going to evolve. It's machine learning, not learning." The models, he says, are getting smarter. They are getting better at sifting data into meaningful signals and meaningless noise. So-called hallucinations, one of the features of all AI models so far, are becoming rarer.
Interestingly, AIEQ's underperformance almost entirely occurred during the two-year slump in the broader stock market that started in late 2021 and ended in late 2023. Amplify switched from daily portfolio rebalancing to monthly in early 2024. Performance has improved.
The WisdomTree AI funds have been operating for about 41/2 years, says Christopher Gannatti, that company's global head of research. He argues that a full test needs longer.
Like most stock-picking funds, these have underperformed the indexes recently because of something particular going on in markets at the moment: the crazy boom in large-cap technology giants. If you simply took profits too early on Nvidia (NVDA) and other "Magnificent Seven" MAGS holdings in recent years, as these and many other active funds have done, the market has made you look foolish - so far.
Stay tuned on that.
Some people are going to make a fortune in the future using AI to pick stocks. There, again, if enough people pick stocks at random, some of them will make a fortune through sheer chance. It is hard to see how any AI model will produce sustained outperformance, because it will simply be copied. The net result may be to make markets more efficient, building an even more compelling case for low-cost index funds.
(Chrissy Bargeron, portfolio manager at financial company Voya, whose AI investment model effectively manages the WisdomTree funds' portfolios, says the key factor for success won't be the AI model but the data.)
But all this lies in the future.
For now, I am reminded of the comment made by "Saki," aka Hector Hugh Munro, a famous English writer from over 100 years ago. "Never be a pioneer," he wrote. "It's the early Christian that gets the fattest lion."
-Brett Arends
This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
July 24, 2026 14:31 ET (18:31 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments