Hugging Face Daily Papers · September 17, 2026 · 4 min read

EvolveTrade: Experience-Driven Policy Refinement for Self-Evolving LLM Trading Agents

Mirrored from Hugging Face Daily Papers for archival readability. Support the source by reading on the original site.

Like Read original ↗

We present EvolveTrade, a framework that lets LLM trading agents refine their own tool-use policies from trading experience. The agent revises how it gathers information, checks signals, and manages risk based on past decisions and outcomes. In a case study, EvolveTrade limited losses during NVIDIA’s 17% single-day drop by keeping exposure low under its refined risk-aware sizing policy, holding just 2.9% in NVDA compared with the static baseline’s 10.7%. Check out our paper for more details!</p>\n","updatedAt":"2026-09-17T03:15:10.886Z","author":{"_id":"64cfa0b9749587dbe01d0079","avatarUrl":"/avatars/93ca0a1d9c5578d052c5af0d4d1a0252.svg","fullname":"Yumin Choi","name":"YuminChoi","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":9,"isUserFollowing":false}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.914976179599762},"editors":["YuminChoi"],"editorAvatarUrls":["/avatars/93ca0a1d9c5578d052c5af0d4d1a0252.svg"],"reactions":[],"isReport":false}}],"primaryEmailConfirmed":false,"paper":{"id":"2609.17632","authors":[{"_id":"6aab59b41d9cc4dec79624a0","name":"Sehee Kim","hidden":false},{"_id":"6aab59b41d9cc4dec79624a1","name":"Yumin Choi","hidden":false},{"_id":"6aab59b41d9cc4dec79624a2","name":"Minki Kang","hidden":false},{"_id":"6aab59b41d9cc4dec79624a3","name":"Sung Ju Hwang","hidden":false}],"publishedAt":"2026-09-15T00:00:00.000Z","submittedOnDailyAt":"2026-09-17T00:00:00.000Z","title":"EvolveTrade: Experience-Driven Policy Refinement for Self-Evolving LLM Trading Agents","submittedOnDailyBy":{"_id":"64cfa0b9749587dbe01d0079","avatarUrl":"/avatars/93ca0a1d9c5578d052c5af0d4d1a0252.svg","isPro":false,"fullname":"Yumin Choi","user":"YuminChoi","type":"user","name":"YuminChoi"},"summary":"Large language model (LLM) trading agents can combine market data, news, and executable analysis, but their behavior is often controlled by static hand-written tool-use policies that are fixed before deployment. This limits their ability to adapt how they gather evidence, invoke tools, verify signals, and manage risk under changing market regimes. We introduce EvolveTrade, a self-evolving framework that treats the system prompt of a tool-using trading agent as a text-parameterized policy. After each update interval, a Policy Agent revises this policy using accumulated decision traces and realized portfolio feedback, while keeping the backbone LLM fixed. The updated policy is then used for the next batch of trading decisions, enabling the agent to refine its information-acquisition and portfolio-construction procedure over time. Experiments across multiple market regimes and two LLM backbones show that EvolveTrade often improves Sharpe Ratio and Cumulative Return over fixed-policy LLM baselines, achieving the improved SR and CR in most evaluated settings. Behavioral analyses further show that self-evolved policies increase code-mediated analysis and activate regime-relevant computations; case-level policy-to-return attributions trace how policy-induced allocation changes contribute to realized return differences. These results suggest that adapting the reusable procedure governing tool use is a key direction for building more robust LLM trading agents.","upvotes":30,"discussionId":"6aab59b41d9cc4dec79624a4","organization":{"_id":"6475760c33192631bad2bb38","name":"kaist-ai","fullname":"KAIST AI","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6469949654873f0043b09c22/aaZFiyXe1qR-Dmy_xq67m.png"}},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"64cfa0b9749587dbe01d0079","avatarUrl":"/avatars/93ca0a1d9c5578d052c5af0d4d1a0252.svg","isPro":false,"fullname":"Yumin Choi","user":"YuminChoi","type":"user"},{"_id":"64b74920fe6a108d03fed767","avatarUrl":"/avatars/a2c05b809c36fa5fab8e1a43b3e67051.svg","isPro":false,"fullname":"Minki Kang","user":"Nardien","type":"user"},{"_id":"6a2da6c8ca070ee12c6e396c","avatarUrl":"/avatars/0355287dcabaa67dbc7f0b10b87451f9.svg","isPro":false,"fullname":"Joe Mama","user":"JoeMama123123123","type":"user"},{"_id":"631c386bc73939ffc0716a37","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1662793811119-noauth.jpeg","isPro":false,"fullname":"SeongWan Kim","user":"idgmatrix","type":"user"},{"_id":"68ccd33453ab64fed782f931","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/23pXpnh_Fn1CuCMShf6eX.png","isPro":false,"fullname":"Sehee Kim","user":"SeheeKim","type":"user"},{"_id":"6a6a9417a569279e390e1343","avatarUrl":"/avatars/4866974d3fab4ba081b094cd0819de75.svg","isPro":false,"fullname":"Mark Davis","user":"silverArcL","type":"user"},{"_id":"6a6aa38fbbca071c718a29de","avatarUrl":"/avatars/93253032d7014e832fa66c1575e1fb9e.svg","isPro":false,"fullname":"James Johnson","user":"Kestrel-Evan","type":"user"},{"_id":"6a6c7c392cdbd2d07b4e26fa","avatarUrl":"/avatars/4097ee3de713cb0bd315c851e81ba635.svg","isPro":false,"fullname":"Michael Garcia","user":"Michael-Garcia","type":"user"},{"_id":"6a6c94df2fbbe664ac69e392","avatarUrl":"/avatars/7c9f0c3490f5cabad92a76f45b60c8bd.svg","isPro":false,"fullname":"Susan Thomas","user":"frostEli","type":"user"},{"_id":"6a6d58512a48f063703b8d3d","avatarUrl":"/avatars/25aa979d4bd497cee90a5429fa13f354.svg","isPro":false,"fullname":"Richard Martinez","user":"Quiet-Richard","type":"user"},{"_id":"6a6c94b74804b5ea70e1608e","avatarUrl":"/avatars/7976fe663fc956d24044fee1356f0049.svg","isPro":false,"fullname":"Joseph Thomas","user":"Azure-Joseph","type":"user"},{"_id":"6a6dc8f808e6705013be79b6","avatarUrl":"/avatars/435f162fa1fc1589195b9c0f231904b5.svg","isPro":false,"fullname":"George Garcia","user":"GeorgeGarcia","type":"user"}],"acceptLanguages":["en"],"dailyPaperRank":0,"organization":{"_id":"6475760c33192631bad2bb38","name":"kaist-ai","fullname":"KAIST AI","avatar":"https://cdn-avatars.huggingface.co/v1/production/uploads/6469949654873f0043b09c22/aaZFiyXe1qR-Dmy_xq67m.png"},"markdownContentUrl":"https://huggingface.co/buckets/huggingchat/papers-content/resolve/2609/2609.17632.md","query":{}}">

Papers

arxiv:2609.17632

EvolveTrade: Experience-Driven Policy Refinement for Self-Evolving LLM Trading Agents

Published on Sep 15

· Submitted by

Yumin Choi on Sep 17

KAIST AI

Upvote

Authors:

Abstract

Large language model (LLM) trading agents can combine market data, news, and executable analysis, but their behavior is often controlled by static hand-written tool-use policies that are fixed before deployment. This limits their ability to adapt how they gather evidence, invoke tools, verify signals, and manage risk under changing market regimes. We introduce EvolveTrade, a self-evolving framework that treats the system prompt of a tool-using trading agent as a text-parameterized policy. After each update interval, a Policy Agent revises this policy using accumulated decision traces and realized portfolio feedback, while keeping the backbone LLM fixed. The updated policy is then used for the next batch of trading decisions, enabling the agent to refine its information-acquisition and portfolio-construction procedure over time. Experiments across multiple market regimes and two LLM backbones show that EvolveTrade often improves Sharpe Ratio and Cumulative Return over fixed-policy LLM baselines, achieving the improved SR and CR in most evaluated settings. Behavioral analyses further show that self-evolved policies increase code-mediated analysis and activate regime-relevant computations; case-level policy-to-return attributions trace how policy-induced allocation changes contribute to realized return differences. These results suggest that adapting the reusable procedure governing tool use is a key direction for building more robust LLM trading agents.

View arXiv page View PDF Add to collection