Should you let a supercomputer pick stocks for your retirement portfolio? Bottom Line Personal asked analyst Neena Mishra, CFA…

More than two dozen exchange-traded funds are powered by AI, ranging from Buy the Dip ETF (DIP) to the Merlyn.AI Bull-Rider Bear-Fighter ETF (WIZ) to AI Powered Equity ETF (AIEQ). Recent studies even suggest that ChatGPT, the AI program that ­creates human-sounding responses, can predict stock price movements more accurately than basic analysis models.

But don’t fire your investment adviser just yet. Using sophisticated algorithms hasn’t delivered for investors. Take AIEQ, the largest ETF in the category, launched back in 2018. It is run by the IBM supercomputer Watson, which has bested world-class chess players and Jeopardy champions. But over the past five years, AIEQ had an annualized return of 4.14% versus 11% for the S&P 500 index. Last year, Watson’s stock picks plunged 32% versus the index’s 19% fall.

Reasons for the poor performance: AI works best sifting through historical data to find patterns and correlations that should lead to profitable trades. But over shorter periods, the market can be irrational, and AI systems struggle to predict unexpected events. Another factor: Supercomputers aren’t cheap—AIEQ charges a hefty 0.75% expense ratio in addition to high trading costs triggered by a 1,700% annual turnover in its portfolio.

Related Articles