JPMorgan's AI beat the 60-40 in tests; advisors aren't worried

Although JPMorgan recently showed that AI can outperform a 60-40 portfolio, investors are still likely to turn to human advisors during economic turmoil.

JPMorgan turned heads last week with a report suggesting that AI investing agents could beat the returns of a standard 60-40 portfolio, which allocates 60% of an investor's assets to stocks and 40% to bonds. Pitting custom-built AI agents against historical data from the past two decades, every hypothetical backtest outperformed a 60-40 allocation. In its best performance, the AI agents beat the 60-40 allocation by 0.7 percentage point a year, with lower volatility.

Still, investors remain wary of consulting AI on portfolio construction, particularly during times of economic uncertainty. The accounting and consulting giant PwC published survey results this week suggesting that in periods of market volatility, investors are much more likely to turn to news articles or financial advisors for guidance.

Only 24% turn to AI in rough markets

Posing the question, "Which of the following tools or resources have you engaged with, or are you considering engaging with, to help make financial decisions during this period of market volatility?" to more than 1,000 respondents, PwC found that only 24% said they would rely on AI-powered tools or assistants. That lagged far behind the 50% who said they would consult online research and financial news and the 48% who said they would go to a financial advisor.

PwC found that the relative distrust of AI comes amid financial service firms' continued search for new ways to use the burgeoning technology. So far, many have employed it to provide answers to research questions, write up summaries of advisors' discussions with clients and automate repetitive back-office tasks. Few, though, have been willing to go so far as to entrust it with investment decisions.

JPMorgan's testing of its custom-built AI agents against a standard 60-40 portfolio seems like a step in that direction. Its analysts were able to use those AI agents to beat not only a 60-40 portfolio but also what it calls its rules-based market regime model. That model evaluates current market conditions like inflation and economic growth and places them in one of four categories: Goldilocks, reflation, stagflation and risk-off.

"The AI agent can be set up with a process to be empowered to make decisions under uncertainty, producing outperformance vs a reasonable benchmark," the group of analysts, led by strategist Thomas Salopek, wrote last week.

What if the AI cheated?

All the same, JPMorgan offered reasons to be skeptical of its apparent success in using AI to outperform a standard 60-40 portfolio. The Salopek-led analysts wrote, "We strongly caution against uncritically accepting what amounts to in-sample, overly confident answers of AI."

"Agentic AI needs to be grounded in a well thought-out asset allocation process, rather than naively assuming the agent can be the source of the domain knowledge," they added.

JPMorgan noted that its tests were conducted by looking at the actual performance of 60-40 portfolios during certain past periods and then hypothesizing how an AI-powered system might have performed if it were investing at the same time. The researchers took steps to prevent the AI from using actual historic market results to influence its decisions.

But there's a chance the systems essentially cheated.

"Although the data is lagged and the prompt is date-anonymized, the LLM models are still trained on data after the cut-off point and may implicitly recall the outcome of recognizable historical episodes (e.g., 2008, COVID, etc.)," according to the analysts' report.

The researchers also noted that heavy reliance on AI-guided trades could come with its own risks, going beyond whether the systems make the "right" investing calls or not. Markets could be distorted, for instance, if investors are crowded into trades that appear well-suited for certain economic conditions.

Advisors have faced tech threats before

For Bryan Byrer, the founder of Millennial Financial Planning in Indianapolis, JPMorgan's successful tests of AI agents against standard 60-40 portfolios are just another technological development to be taken in stride. He said technical advances of one form or another always seem to be coming for advisors' jobs.

He has yet to see a client use AI to second-guess his financial advice, he said. Even if one did, he's far from perceiving AI as an imminent threat to his business.

"Intelligence is different from emotions, and people do emotional things with money but not always intelligent things with money," Byrer said. "That's an understanding that we won't ever get, or at least for a very long time, from AI."