JPMorgan AI Tops 60/40 Portfolio by 0.7% in Two-Decade Backtest

Yahoo Finance ·

JPMorgan AI Tops 60/40 Portfolio by 0.7% in Two-Decade Backtest Khac Phu Nguyen Fri, July 10, 2026 at 3:44 PM EDT 2 min read JPM This article first appeared on GuruFocus . JPMorgan Chase ( NYSE:JPM ), a major Wall Street bank, has been testing whether artificial intelligence can move beyond supporting investors and begin making asset-allocation decisions itself. Researchers at the bank developed eight AI-powered investment agents that shifted money between stocks and bonds as market conditions changed. In backtests covering the past two decades, the strongest system outperformed a traditional portfolio holding 60% stocks and 40% bonds by 0.7 percentage point annually while also delivering lower volatility. JPMorgan strategists led by Thomas Salopek described the project as the bank's first effort to build an AI system capable of identifying different market regimes. Warning! GuruFocus has detected 9 Warning Signs with JPM. Is JPM fairly valued? Test your thesis with our free DCF calculator. The agents were powered by models from OpenAI, an artificial intelligence developer, and Anthropic, an artificial intelligence company, and were designed to classify markets into four environments: Goldilocks, reflation, stagflation and risk-off. The systems then adjusted allocations across asset classes, favoring equities during periods of stronger growth and increasing fixed-income exposure as economic conditions weakened. All eight agents outperformed the conventional 60/40 portfolio on a risk-adjusted basis and also surpassed JPMorgan's existing rules-based market-regime model. These results may suggest that AI could improve investment frameworks already used to guide portfolio allocation, although the findings were based entirely on historical simulations rather than live trading. The experiment may offer investors an early look at the next stage of Wall Street's AI adoption, as financial institutions test whether large language models can progress from research, coding and internal tools toward direct capital-allocation decisions. However, JPMorgan warned investors against viewing the backtests as evidence that AI can consistently outperform markets, noting that overly confident in-sample results can be misleading. The bank's strategists also acknowledged that wider use of similar AI models could contribute to crowded trades, make markets more vulnerable to manipulation and amplify stress if multiple firms reach the same conclusions. While the researchers expressed enthusiasm about agentic AI, they remained cautious about handing full asset-allocation authority to a model and argued that these systems should operate within a carefully structured investment process.

DYAX Investor Sentiment

Bullish (Long) 38% · Bearish (Short) 62%

434 participants

Related News

원문 보기 — Yahoo Finance