JPMorgan Chase & Co. is testing artificial-intelligence “agents” designed to allocate capital between stocks using market and economic data, according to multiple reports. Bloomberg says the bank’s strategists are assessing whether an AI model can perform beyond conventional portfolio approaches by making allocation decisions itself rather than relying on fixed rules. The outlets also report that JPMorgan tested multiple versions of the agents in historical simulations and that the agents outperformed a traditional 60/40 portfolio, which typically divides assets between stocks and bonds.
PYMNTS and Bloomberg add that the bank compared the AI agents not only against the 60/40 benchmark but also against JPMorgan’s own rules-based market regime model. Several sources characterize the work as an early look at how Wall Street may adopt AI for investment decision-making and risk management.
MarketWatch and other outlets note there is “good and bad news,” reflecting that the findings are based on backtests and simulations rather than real-world performance. JPMorgan also frames the approach as potentially involving less risk, but the reports emphasize the results are not live, audited track records.
