Cross Asset Systematic Highlights: With a Little Help from My Friends: Agentic AI for Market Regimes and Asset Allocation

研报英文原文证据摘录

Cross Asset Systematic Highlights: With a Little Help from My Friends: Agentic AI for Market Regimes and Asset Allocation

J P M O R G A N Global Markets Strategy

09 July 2026

Cross Asset Systematic

Highlights

With a Little Help from My Friends: Agentic AI for

Market Regimes and Asset Allocation

• We are enthusiastic about the possibilities of agentic AI, even as we are wary Cross Asset Systematic Strategy

AC to hand off asset allocation decision-making to an agent. Thomas Salopek

• Given that we know how different asset classes behave in different market (1-212)thomas.salopek@jpmorgan.com834-5476

regimes, we ask how well an AI agent can do identifying the current macro

J.P. Morgan Securities LLC

regime. AC Sandra Djambazovska

sandra.djambazovska@jpmorgan.com• 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 J.P. Morgan Securities plc

thought-out asset allocation process, rather than naively assuming the agent Dobromir Tzotchev, PhD

can be the source of the domain knowledge. (44-20) 7134-5331

dobromir.tzotchev@jpmorgan.com

• We quantify the degree to which our AI agent is in-sample or unfairly looking J.P. Morgan Securities plc

ahead, and we offer corrections to mitigate potential cheating in the results.

• Overall, we are pleased with the result, meaning the AI agent can be set up with

a process to be empowered to make decisions under uncertainty, producing

outperformance vs a reasonable benchmark. Here the specific application was

market regime identification, but so long as there is a deterministic process for

our decision-making, we can give the agent the advice it needs. And we can

broaden our agentic efforts to other asset allocation questions.