{"paper":{"title":"The Alpha Illusion: Reported Alpha from LLM Trading Agents Should Not Be Treated as Deployment Evidence","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"Reported alpha from LLM trading agents should not be treated as deployment evidence.","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.CE","authors_text":"Ao Hu, Danilo Mandic, Danny Dongning Sun, Juncheng Bu, Jun Han, Liangjian Wen, Xu Yinghui, Yiyi Chen, Yuxuan Ye, Zenglin Xu","submitted_at":"2026-05-16T09:14:35Z","abstract_excerpt":"End-to-end LLM trading agents have moved quickly from research curiosity to a small ecosystem of named systems, including FinCon, FinMem, TradingAgents, FinAgent, QuantAgent, and FLAG-Trader. Several of these report headline Sharpe ratios that would be material if read at face value on a deployment desk, and associated benchmarks such as FinBen report trading-task Sharpe statistics in the same range. The gap between architecture research and deployment claim has been crossed too freely on both sides of the academia--industry divide. We take a position on that gap: reported alpha from end-to-en"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Before such returns can support claims of deployable trading capability, they must survive structural validity tests for temporal integrity, real-world frictions, counterfactual robustness, predictive calibration, numerical execution, and multi-agent disaggregation.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"Current public evidence cannot yet distinguish robust predictive ability from temporal contamination, unmodeled frictions, short-window Sharpe uncertainty, narrative fitting, and parametric priors (abstract, paragraph on the gap between architecture research and deployment claim).","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Reported alpha from end-to-end LLM trading agents does not constitute deployment evidence until it passes structural tests for temporal integrity, frictions, robustness, calibration, execution, and disaggregation.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Reported alpha from LLM trading agents should not be treated as deployment evidence.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"4bf44e70acb785848c5c16a08de3f4a672a4096c901f3d23298170ea2a63b80c"},"source":{"id":"2605.16895","kind":"arxiv","version":1},"verdict":{"id":"aa57d4df-d0d7-4936-91e8-35a007afee9d","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T19:14:14.265450Z","strongest_claim":"Before such returns can support claims of deployable trading capability, they must survive structural validity tests for temporal integrity, real-world frictions, counterfactual robustness, predictive calibration, numerical execution, and multi-agent disaggregation.","one_line_summary":"Reported alpha from end-to-end LLM trading agents does not constitute deployment evidence until it passes structural tests for temporal integrity, frictions, robustness, calibration, execution, and disaggregation.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"Current public evidence cannot yet distinguish robust predictive ability from temporal contamination, unmodeled frictions, short-window Sharpe uncertainty, narrative fitting, and parametric priors (abstract, paragraph on the gap between architecture research and deployment claim).","pith_extraction_headline":"Reported alpha from LLM trading agents should not be treated as deployment evidence."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.16895/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"cited_work_retraction","ran_at":"2026-05-19T20:52:41.311578Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T19:31:18.958633Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T19:20:48.757248Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T18:41:56.280208Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T18:33:26.358915Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"2292a4b8a828b65ced0922ca122c707ce3e73a452995539d07ff323d40893035"},"references":{"count":41,"sample":[{"doi":"","year":2000,"title":"Optimal execution of portfolio transactions.Journal of Risk, 3(2):5–39, 2000","work_id":"5f827f4c-330b-4c1c-b60f-2e87e5b8280c","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Stockbench: Can llm agents trade stocks profitably in real-world markets?","work_id":"df5e4008-d3f7-4be2-be64-782b28c46bb2","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"Empirical asset pricing via machine learning.The Review of Financial Studies, 33(5):2223–2273, 2020","work_id":"b351bc77-e673-460c-97cb-b66251e9ace3","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2017,"title":"Chuan Guo, Geoff Pleiss, Yu Sun, and Kilian Q. Weinberger. On calibration of modern neural networks. InInternational Conference on Machine Learning (ICML), pages 1321–1330. PMLR, 2017","work_id":"f00102f9-22d3-4720-bca5-731df17ea507","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2016,"title":"Harvey, Yan Liu, and Heqing Zhu","work_id":"ef8bbdf8-ee5a-4069-a48d-12d40963b996","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":41,"snapshot_sha256":"0c19b4ee29f8a62761112a4c6735e5c1407345c7f2786f42876d085f9fc10ad0","internal_anchors":4},"formal_canon":{"evidence_count":1,"snapshot_sha256":"951e6379b61a611827fe8ed39a01d83378fe642fed0a1772c853c40a8c16b4f9"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}