Breaking

JPMorgan Builds AI Agents for Asset Allocation, Topping 60/40 in Tests

July 9, 2026 at 22:45 EDT

AI Agents
Finance

A research team at JPMorgan Chase & Co. has built AI agents that autonomously shift capital between stocks and bonds, and in internal testing they outperformed the traditional "60/40" portfolio. A team led by strategist Thomas Salopek developed an array of AI-powered investing agents that dynamically adjust the allocation between equities and bonds as market conditions change, running a backtest over the past 20 years of data.

JPMorgan Chase · AI Asset Allocation

AI Agents Beat the Classic 60/40 Portfolio in Backtests

A JPMorgan research team built autonomous AI agents that shift capital between stocks and bonds as markets move. Over a 20-year backtest, the best system topped the traditional 60% stocks / 40% bonds strategy — with lower volatility.

+0.7 pp

Annual outperformance vs. the 60/40 portfolio

20 yrs

Historical data covered in the backtest

Lower

Volatility than the classic 60/40 mix

Annualized return, backtested

Same unit — return per year, shown to scale (60/40 baseline = reference)

Baseline

60/40

Traditional 60% stocks / 40% bonds

→

+0.7 pp / yr

+0.7

AI Agent

Autonomous AI allocation

The AI system also beat JPMorgan's existing rule-based market-regime model.

How the agents work

Read changing market conditions

→

Decide stock / bond mix autonomously

→

Shift capital as markets move

Why it matters

JPMorgan spends ~$2B/year on AI and has moved from stock picking to the bigger question: can AI be trusted with whole-portfolio allocation? Rivals including Goldman Sachs are exploring the same.

Still untested

Results come from historical backtests only — no real-money deployment announced. Behavior in live markets and resilience in unexpected conditions remain open questions.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.