LLMs in Finance
Where language models are used in financial work, what the papers and filings show, and what they get wrong.
-
Language models reading financial statements: published accuracy
Four 2026 benchmarks measured LLMs computing ratios, verifying statements, answering from full 10-Ks, and summarising MD&A. Accuracies and failure modes, cited.
-
Time-series forecasting with LLMs: claims, benchmarks, and sceptics
What 2026 papers report when language and time-series foundation models forecast returns, volatility, and factor rankings against simple baselines, by arXiv id.
-
Extracting tables from filings: a model versus XBRL, when each wins
XBRL company facts for one filer, deduplicated, with the tag changes and restatements any extractor hits, and what 2026 table-extraction benchmarks report.
-
Earnings-call sentiment: what the papers measured, FinBERT to LLMs
What financial-sentiment research reports, from the 2019 FinBERT paper to 2026 studies of call tone, evasion, KPI extraction, and lookahead bias, by arXiv id.
-
What companies say about AI in 10-K risk factors: an EDGAR read
10-K documents matching four AI phrases per year since 2019 from EDGAR full-text search, a 40-filing sample with accession numbers, and what the counts cannot say.