Dataset

SEC-EDGAR

Datamule partnered with Teraflop AI and Eventual to publish the SEC-EDGAR dataset, which covers all major filings from the SEC EDGAR database and amounts to 590 GB across about 8 million samples and 43 billion tokens. Collection relied on the datamule-python library and John Friedman's official datamule API, tools designed in Python for gathering and handling SEC filing data.

Open on Hugging Face

Data preview

Subset

Loading preview...

Dataset card

SEC-EDGAR is a large public dataset of corporate filings scraped from the U.S. Securities and Exchange Commission's EDGAR system. It is distributed via the kapilrao listing on a model-data platform.

What this dataset contains

The dataset packages roughly 590 GB of material covering about 8 million individual filings and 43 billion tokens. It draws from all major filing types filed through EDGAR. For each filing it provides the raw document contents, an extracted plaintext version, and associated metadata.

Structure and fields

Each filing record is accompanied by document-level fields including file type, sequence, filename, description, and SEC SGML byte count. Filing-level metadata includes accession number, filing date, period, document list, and filer attributes such as company name and Central Index Key (CIK). The schema therefore spans both the textual payloads and the structured identifiers used by EDGAR itself.

How analysts use it

The corpus is well suited to training or evaluating text-generation and text-classification models on regulatory disclosures. Investment analysts and quants can use it to build NLP pipelines over 10-K, 10-Q, and other filings for feature engineering, summarization, or sentiment extraction. The presence of accession numbers, CIKs, and period fields also enables event-study and time-series joins against market data.

Documentation & references 1