FinObservatory

Research library

The text substrate of the grounded analyst

FinObservatory’s research library is the acquired text corpus behind the future grounded analyst: what central banks said, decided, and reported, on disk, cataloged, and verifiable. Four corpora, 39,467 documents, 1936 to 2026. This module explores the catalog, the unified metadata index; document text stays on local disk under each publisher’s terms and every entry links out to the publisher’s own copy where one exists.

39,467
Documents in the catalog
4
Corpora
1936–2026
Document span
71.8 GB
Text on local disk
What this shows, and what it never will. Fed and SEC materials are US government works in the public domain. BIS speeches and IMF reports are not: those corpora live on local disk for research grounding only, and these pages show metadata (title, author, date) plus a link to the publisher, never the text. The full license posture, per corpus and with the recorded license language, is on the methodology page.

Corpus inventory

Every count and span below is computed from the catalog at render time. Two corpora have dedicated explorers: FOMC materials and central-banker speeches.

CorpusDocumentsSpanOn diskSourceLicense posture
BIS central bankers' speeches
bis_speeches
20,728199620260.4 GBBank for International Settlements, bis.org/cbspeechesNoncommercial local-use. BIS terms: noncommercial use; extracts over 400 words need permission. Metadata and links here; full text at bis.org only.
SEC EDGAR 10-K / 10-Q filings, US financial firms
edgar_filings
9,8462016202653.8 GBUS SEC, EDGAR full-text archivesPublic domain. US government work, public domain.
Federal Reserve FOMC and Board materials
fed_text
4,691193620263.0 GBBoard of Governors, federalreserve.govPublic domain. US government work, public domain (Board seals and logos excluded).
IMF Article IV staff reports and GFSR
imf_reports
4,2022002202614.6 GBIMF eLibrary, elibrary.imf.orgLocal-use only. IMF copyright: personal, noncommercial usage only; no redistribution. Metadata and eLibrary links here; PDFs stay on local disk.

Source: FinObservatory text catalog (text_catalog.parquet), built from the four raw corpora IMF documents carry no per-document date in the eLibrary catalog; the IMF span uses the report year (imf_year). On-disk bytes are file sizes, except BIS rows where bytes measure the speech text inside the single bulk csv. Methodology

Documents per decade

The catalog by decade of document date (report year for IMF). The long left tail is the Federal Reserve’s historical minutes and their predecessor records, which reach back to the 1930s; the modern bulk arrives after 1996, when the BIS speeches archive and the Fed’s web-era publications begin, and after 2016, the EDGAR filing floor.

BIS central bankers' speechesFederal Reserve FOMC and Board materialsSEC EDGAR 10-K / 10-Q filings, US financial firmsIMF Article IV staff reports and GFSR
05,00010,00015,00020,0001930s1940s1950s1960s1970s1980s1990s2000s2010s2020s

Source: FinObservatory text catalog, grouped by decade and corpus Methodology

Most recent documents

The newest documents in the catalog by document date. IMF reports cannot appear here: their catalog records year and issue but no document date, so ranking them by date would be invented precision.

Source: FinObservatory text catalog, newest document dates Links go to the publisher's own copy: bis.org for BIS speeches, federalreserve.gov for Fed materials, sec.gov filing indexes for EDGAR. Methodology

Explore: FOMC materials | central-banker speeches | acquisition and licensing methodology.