JPMorgan tests AI agents for stock and bond allocation

Traders Union
All News
Financial Institutions
JPMorgan tests AI agents

Highlights

JPMorgan’s AI agent beat 60/40 by 0.7 percentage point a year in backtests.
All eight agents outperformed on a risk-adjusted basis.
The bank says live-market results remain unproven.

Researchers led by strategist Thomas Salopek designed a group of AI agents that shift between stocks and bonds as market conditions change, Bloomberg reported. The best-performing system beat a classic 60/40 portfolio by 0.7 percentage point a year over two decades of simulations, while also producing lower volatility and outperforming JPMorgan’s own rules-based market regime model.

AI moves from research tool to allocator

The experiment marks a step beyond the way banks have mostly used large language models so far. Over the past two years, Wall Street firms have embedded AI into research, coding, client tools, and internal analysis. JPMorgan’s test asks a more consequential question: whether AI can help decide how capital is split across markets.

The agents were built using models from OpenAI and Anthropic. They classified markets into four regimes based on growth and inflation: Goldilocks, reflation, stagflation, and risk-off. From there, they adjusted allocations across asset classes, favoring equities when growth was strong and increasing bond exposure when conditions weakened.

All eight AI agents tested outperformed the 60/40 portfolio on a risk-adjusted basis. They also beat the bank’s existing regime-based framework, suggesting the systems found useful patterns in historical market environments.

Backtests come with limits

JPMorgan’s strategists warned against giving too much weight to the results. Backtests can look strong because they are built on known historical data, and AI systems can produce answers that appear more confident than the evidence justifies.

That caveat matters because the wider adoption of similar models could create its own risks. If many firms rely on comparable AI systems, trades may become more crowded, markets may react faster to the same signals, and stress periods could become more amplified.

The next test for AI on Wall Street

The study matters because asset allocation sits at the center of investment management. If AI can reliably read market regimes and adjust portfolios, it could become a serious tool for large banks, pension funds, and wealth managers.

But the hurdle is high. JPMorgan’s own warning is the key point: agentic AI may help structure decisions, but it still needs human oversight, a disciplined investment process, and live-market proof before it can be trusted with capital allocation at scale.

Earlier, we reported that JPMorgan and Ripple executed the first instant treasury settlement on XRP Ledger.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.