{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,15]],"date-time":"2026-10-15T08:44:13Z","timestamp":1792053853739,"version":"4.4.0"},"posted":{"date-parts":[[2026]]},"group-title":"SSRN","reference-count":0,"publisher":"Elsevier BV","license":[{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"unspecified","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"abstract":"<jats:p>Large language models (LLMs) now drive trading systems, from signal extractors to autonomous agents, but the literature is fragmented across markets, architectures, and evaluation practices. This survey traces the path from quantitative trading to LLM agents, organizes the literature by capability (perception, reasoning, action, memory, collaboration) and by six markets from equities to prediction markets, and maps 47 systems across both to show where evidence is thin. We audit evaluation protocols and consolidate head-to-head evidence against traditional quantitative baselines. Four findings emerge. First, the margins of independently rerun agents shrink or vanish, though the reruns cover few systems and many reported margins rest on windows of eight months or less, where an annualized Sharpe ratio has a standard error above one. Second, the best-supported hypothesis places the LLM upstream of the trade (text signals, factor and model code, and bounded execution tilts), although most upstream evidence comes from single author-run studies. Third, controlled comparisons do not establish a multi-agent advantage, though risk controls help where ablated. Fourth, structure and input design matter alongside the backbone, and the best structure depends on the backbone.<\/jats:p>","DOI":"10.2139\/ssrn.7610285","type":"posted-content","created":{"date-parts":[[2026,10,15]],"date-time":"2026-10-15T07:47:08Z","timestamp":1792050428000},"source":"Crossref","is-referenced-by-count":0,"title":["From Algorithms to Agents: A Survey of LLM-Based Trading Systems"],"prefix":"10.2139","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-7235-2377","authenticated-orcid":true,"given":"Yebo","family":"Feng","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yichen","family":"Luo","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jingyang","family":"Dai","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Chien-Chih","family":"Chen","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Guangyan","family":"Gan","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3993-5263","authenticated-orcid":true,"given":"Jiahua","family":"Xu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lin William","family":"Cong","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yang","family":"Liu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"78","container-title":[],"original-title":[],"deposited":{"date-parts":[[2026,10,15]],"date-time":"2026-10-15T07:47:08Z","timestamp":1792050428000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.ssrn.com\/abstract=7610285"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":0,"URL":"https:\/\/doi.org\/10.2139\/ssrn.7610285","relation":{},"subject":[],"published":{"date-parts":[[2026]]},"subtype":"preprint"}}