SSRNML & AI Methods
Short Windows, Unmodelled Costs, Unknown Cutoffs: A Pre-Registered Evidence Audit of 91 LLM Trading Papers
Pre-registers and audits 91 LLM trading papers, finding 51% ignore trading costs, median test window is 12 months, and 98% fail to control for look-ahead bias in model training.
Featured in No. 139 on 13 Nov 2026 · on release day
- Released
- 13 Nov 2026
- First featured
- No. 139 · 13 Nov 2026
- Published in
- Not yet, as far as Semantic Scholar knows
- Fanfare
- 4 of 5
- Identifier
- SSRN 7687481
- Authors
- Heng Zhang
Citations and venue from Semantic Scholar (ODC-BY), refreshed weekly. Summary: Quant Letter (CC BY 4.0).