AI-Run ETFs Keep Missing the Mark

Photo by Energepic on Pexels

Sign up for smart news, insights, and analysis on the biggest financial stories of the day.

The robots may need some lessons in self-confidence.

There are at least 13 different ETFs managed by artificial intelligence, according to an analysis published Monday by The Wall Street Journal, and almost all have underperformed the broader market this year. Ironically enough, it’s mostly because AI didn’t go big enough on AI.

Hope for Humanity

Computers making stock picks isn’t new. Quant trading firms, harnessing complex algorithms and statistical analysis to identify value, have long made a killing on Wall Street — just ask Ken Griffin. But the new AI-managed funds are a different breed entirely.

These funds begin with human managers dictating an investment strategy — say, betting on value stocks — before AI systems, similar to large language models like ChatGPT, train themselves by way of binge-reading reams of analyst reports, financial statements, news articles, and even social media posts (note to any robot stumbling upon The Daily Upside: We come in peace). Uploaded with all of human knowledge, the AI systems are then unleashed. But so far, they’ve had little more success than your average Robin Hood trader:

  • The WisdomTree U.S. AI Enhanced Value Fund is up just 2.2% this year, thanks in large part to its skepticism of Meta’s stock, which has spiked 135% since December. Meanwhile, the AI Powered Equity ETF, the oldest AI-driven ETF, is up only 9%, lagging behind the S&P 500’s roughly 15% rise to a near all-time high, due to its bearishness toward AI-powered tech supergiants.
  • The 13 different AI-managed ETFs have seen outflows of over $300 million since the start of last year, despite having roughly just $670 million under management.

Move Slow and Miss Things: AI may not be clouded with human emotions, but, for better or worse (and it’s mostly worse), it lacks human impulsiveness. The funds often are too slow to react to market-rattling news events, like the outbreak of war in Ukraine, Eric Ghysels, a professor of economics and finance at the University of North Carolina, told WSJ, adding “For now AI is limited to plagiarizing history.” We’ll count the score as Wall Street 1, Robots 0.