The full COT history, in one download.
Every market. Every Friday. Every byte.
+380 CFTC markets, every report family, from 1986 to today — delivered as CSV and NDJSON in a single, schema-versioned archive.
COT Data Dump
The COT Data Dump is a one-time purchase that gives you the entire CFTC Commitments of Traders historical dataset — all four report families (Legacy, Disaggregated, Traders in Financial Futures, Supplemental), every market the CFTC has ever published, in both CSV and NDJSON. The underlying data is U.S. federal public domain (17 USC §105); the value of this product is the curation, packaging, and SHA-256 verification.
cot-data-dump.zip ├── cot-futures-only.csv Weekly bars · 1986 → today ├── cot-futures-only.ndjson same data, NDJSON ├── cot-disaggregated.csv Weekly bars · 2006 → today ├── cot-disaggregated.ndjson ├── cot-tff.csv Weekly bars · 2006 → today ├── cot-tff.ndjson ├── cot-supplemental.csv Weekly bars · 2006 → today ├── cot-supplemental.ndjson ├── markets.csv Every CFTC market · code → category ├── COLUMNS.md Per-column reference for every field ├── manifest.json SHA-256 per file + row counts └── README.md
What's inside
Four CFTC families
Legacy futures-only, Disaggregated F&O Combined, Traders in Financial Futures, and Supplemental — all in the same archive.
CSV + NDJSON
Same data, both formats. CSV for spreadsheets and analysts, NDJSON for streaming pipelines and engineers.
Markets dictionary
A markets.csv joins every CFTC code to category, exchange, contract units, and the report families it belongs to.
Verifiable manifest
manifest.json ships SHA-256 checksums for every file plus row counts, so you know what you got matches what we packed.
Schema-versioned, lifetime-stable
Every archive is tagged v1, v2, etc. Adding a column is non-breaking — schema stays. We only bump the version when CSV columns rename, remove, or reorder, and active customers are notified before any breaking change. Pin your Power Query / Pandas pipeline to a version with confidence.
You own what you download
The CFTC publishes Commitments of Traders data as part of its regulatory mandate. U.S. federal works are public domain under 17 USC §105 — no copyright applies. You can share, redistribute, build paid products on top, or feed it into any model. We assert no rights in the underlying data; the value of this product is the curation, the schema discipline, and the verification.
Built for
- Quants backtesting positioning strategies
- Research desks building factor models
- Newsletters that need clean weekly numbers
- Analysts tired of scraping the CFTC website
- Engineers integrating COT into custom dashboards
- AI/ML teams training positioning-aware models
Ship faster than scraping yourself.
Save the week of plumbing, the SODA pagination edge cases, the column-name typos in the Supplemental dataset. Buy once, never look at the CFTC website again.
Instant download. 30-day re-download window. Secure Stripe checkout.
Frequently asked questions
What exactly do I get?
A single ZIP (`cot-data-dump.zip`) containing four CSVs and four NDJSONs — one pair per CFTC report family — plus `markets.csv` (every CFTC market with code, exchange and category), `COLUMNS.md` (per-column reference for every field across all four families with examples), `README.md` (overview + how to verify), and `manifest.json` (schema version, asOfDate, SHA-256 hash + row count for every file). About +380 markets, weekly history back to 1986 for the Legacy family and 2006 for Disaggregated / TFF / Supplemental.
Is the data really free elsewhere?
Yes. CFTC publishes the underlying data on its website as U.S. federal public domain. What you're paying for is the work of pulling every dataset, normalising column-name typos (the CFTC ships 'postions' instead of 'positions' in the Supplemental file), packaging it cleanly, and shipping a verifiable manifest.
How often is the archive updated?
The archive refreshes automatically each Friday within minutes of CFTC's 15:30 ET release. If CFTC delays a release (US holiday weeks, technical shifts), our watcher polls every hour in a 4-hour window covering both Eastern Daylight and Standard Time — the freshest data lands as soon as CFTC publishes, never weeks later. Every download in your 30-day window returns the latest archive, not a snapshot from your purchase date.
Can I redistribute or build products on top?
Yes — the CFTC data itself is public domain, no licence required. We don't restrict what you do with the bytes. We do ask that you don't redistribute our schema-versioned ZIP file as-is for free, since that undercuts our packaging work; build something on top instead.
What about refunds?
EU/UK customers have the standard 14-day cooling-off right unless the download has started. After download, the product is non-refundable (digital goods, performance has begun). Outside that, contact support@cot-reports.com for any issue.
Will the schema break my pipeline?
Not without warning. The current `schemaVersion: "v1"` in `manifest.json` is a contract: adding columns stays on v1 (non-breaking). Renaming, removing or reordering columns bumps to v2 and is announced on the changelog before it lands. Pin to the version in the manifest, build your parser once, and trust the contract.
Can I open the files in Excel or Google Sheets?
Yes — every CSV opens natively in Excel, Google Sheets, Numbers, LibreOffice Calc, or any spreadsheet tool. The files are plain UTF-8 with comma separators and quoted text fields. The largest CSV (cot-futures-only.csv) has roughly 280,000 rows, well under Excel's 1M-row sheet limit. For Python or R, use pandas.read_csv() / read.csv() — column types are inferable.
Can I use this for backtesting CFTC-based strategies?
Yes. The archive includes the complete weekly Tuesday-snapshot series for every market across all four CFTC report families. Legacy futures-only goes back to 1986; Disaggregated, TFF, and Supplemental go back to 2006. You get every trader category (Non-Commercial, Commercial, Managed Money, Producer/Merchant, Swap Dealers, Asset Manager, Leveraged Funds, Dealer, etc.) with positions, changes, and percentage-of-open-interest. Pair with daily OHLCV from your price feed for full backtests.
How big is the download and how long does it take?
The compressed ZIP is approximately 100 MB. Uncompressed, the 12 files total around 700 MB — most of that volume is in the NDJSON files (each row is a complete JSON object, more verbose than the equivalent CSV row). On a 50 Mbps connection the download takes ~15 seconds; on slower connections proportionally longer. Delivered via a single signed Storage URL — one download, one file.
What's the difference between the CSV and NDJSON files?
Identical data in two formats. CSV is flat — comma-separated, one row per observation, human-readable in spreadsheets and editors, easy to load into pandas, R, or any database with a CSV importer. NDJSON (newline-delimited JSON) is one JSON object per line, ideal for streaming ingestion into pandas (read_json(lines=True)), Spark, ClickHouse, BigQuery, or any tool that consumes JSON natively. Pick the format that matches your toolchain; you don't need both.