Historical Thoroughbred Racing & Lottery for AI Training (Pre-1930)
Documentation

Dataset Overview
This archive contains historical publications covering Thoroughbred racing, breeding, and lottery economics spanning 1821 to 1930. It is designed for AI training, quantitative research, and historical analysis.

Publications Included

  • Daily Racing Form (1894–1930): 619,998 records across 5,670 issues

  • American Turf Register (1829–1844): 7,690 records across 3,322 issues

  • Cohen’s Gazette & Lottery Register (1821+): 2,534 records across 943 issues

  • Blood-Horse (1929–1931): 2,056 records across 56 issues

Total: 632,000 records across 9,991 unique issues

File Format

  • All records are delivered in JSONL format.

  • Each record contains: issue, title, author, type, text, text_length

  • Manifest files provide an index of every article within an issue.

  • A secure parsed view is available for direct SQL querying.

Processing Standards

  • Professional OCR with quality validation

  • Aggressive cleaning of OCR artifacts

  • Paragraph reconstruction applied

  • Bias-audited where appropriate

  • Full provenance tracking

Provenance & Licensing

  • Source material is pre-1930 and verified public domain.

  • All datasets are copyright-cleared and ethically sourced.

  • Each record includes a provenance statement and bias audit notice.

Contact
For questions or custom licensing, contact: hello(at)devinmediacorp.com

Scroll to Top