Tian Zhang, Meng Li · 2026 · arXiv
DOI: 10.48550/arXiv.2610.26662 · open-access PDF
Agen perdagangan LLM dievaluasi hampir secara eksklusif dalam backtests dan benchmark harnesses otonom. Pengembangan, di mana tindakan irreversible, menimbulkan pertanyaan yang berbeda: siapa yang memverifikasi tindakan agen, dan kegagalan mana yang dapat ditangkap setiap verifier secara struktural? Kami melaporkan pada AllPath, agen perdagangan LLM pribadi yang didistribusikan melawan akun broker kertas hidup. Sistem memantau posisi melalui mesin aturan deterministik, dan setiap malam ia meninjau hari sendiri dan mengusulkan revisi untuk strategi perdagangannya sendiri. Verifikasi dibagi ke dalam tiga lapisan. Kode memverifikasi apa yang seharusnya tidak tergantung.
Machine translation (Meta M2M100); the English original follows.
LLM trading agents are evaluated almost exclusively in backtests and autonomous benchmark harnesses. Deployment, where actions are irreversible, raises a different question: who verifies the agent's actions, and which failures can each verifier structurally catch? We report on AllPath, a personal LLM trading agent deployed against a live paper-brokerage account. The system watches positions through a deterministic rules engine, and each night it reviews its own day and proposes revisions to its own trading strategies. Verification is split across three layers. Code verifies what must not depen
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.26662, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange