Cftc Cot Historical Data: Dataset Structure and Field Coverage

The CFTC Commitments of Traders (COT) Reports break exchange-traded futures and options open interest into trader categories — one row per contract market per report week, further split by maturity bucket. Four report families form the archive:

The Legacy series starts on January 15, 1986, giving nearly four decades of weekly observations. The supplemental CIT report reaches back to January 3, 2006, and the Disaggregated and Traders in Financial Futures families both begin June 13, 2006 — so any backtest that mixes families effectively starts from 2006 unless it stays within Legacy categories.

The PRE API addresses each family through a fixed dataset ID, and the legacy text-file names line up with them one-to-one:

Every row names its market in market_and_exchange_names (for example "WHEAT-SRW - CHICAGO BOARD OF TRADE"), stamps the Tuesday open-interest snapshot in report_date_as_yyyy_mm_dd, and keys the contract with cftc_contract_market_code (001602), cftc_market_code (CBT) and cftc_commodity_code / commodity_name (001 / WHEAT).

Position columns follow: open_interest_all totals all maturities; noncomm_positions_long_all, _short_all and _spread_all carry Legacy speculation; comm_positions_long_all and _short_all carry commercials; nonrept_positions_* cover traders below reporting thresholds; and change_in_open_interest_all gives the week-over-week move. Shares arrive precomputed in pct_of_open_interest_all and related columns, and traders_tot_all counts distinct reporting traders.

The long-format Legacy layout adds old-crop and other-maturity column sets (_old / _other) repeating every position category, which is why Legacy alone can carry maturity breakdowns that TFF cannot. Because shares arrive precomputed as percentages, you can rank crowding across all reporting markets without reconstructing denominators from raw positions.

Sentiment and crowding factors are the standard outputs, built from the Legacy, Disaggregated, TFF and CIT series. The columns themselves dictate the constructions: net position spreads from noncomm_positions_long_all minus _short_all; category concentration from the pct_of_oi_* share columns; participation breadth from the traders_* counts; and momentum of positioning from change_in_open_interest_all.

Timing matters because the report lands Fridays at 3:30 pm ET and reflects the prior Tuesday's open interest, shifting for holidays — that gap sets the earliest timestamp a signal can trade on. Coverage spans CME, CBOT, COMEX, ICE and Coinbase Derivatives markets, so one schema covers energies, metals, grains, rates and crypto futures.

Signals get sharper when positioning joins prices and benchmarks: the EOD Historical Data APIs ship 30-plus years of history across 150,000-plus tickers, while the World Federation of Exchanges Statistics track derivatives volumes for 60-plus exchanges back to February 1970. More workflows sit on the quants using diversified capital markets data page.

Four structural quirks shape any panel you assemble. Markets drop out of reporting entirely when fewer than 20 reportable large traders remain, so series disappear instead of printing zeros. Trader counts are suppressed wherever fewer than four active traders exist in a category. TFF publishes long format only, so maturity breakdowns live solely in Legacy. And CIT extends to just 13 select agricultural contracts.

Timing adds two more: the Tuesday-snapshot-versus-Friday-release lag, and holiday shifts in publication day — a Friday holiday moves release to Thursday or the next business day, so ingestion jobs keyed to Fridays need a calendar check rather than a fixed weekday. If you need bank-level futures exposure rather than trader-category aggregates, the CFTC Bank Participation Reports cover that slice and are cataloged separately under diversified financial services.

None of these are defects; they are how a surveillance program is designed. The CFTC deliberately leaves interpretation to you, which is precisely why the raw columns arrive unrevised and the thresholds that remove markets are stated outright rather than hidden in a methodology appendix.

For the full landscape, read the diversified capital markets data guide, which maps all ten pooled datasets in the industry. Browse the diversified-capital-markets data hub for every cataloged source, rank them on the best diversified capital markets datasets list, or compare API-access options on the diversified capital markets APIs page. If your work extends from positioning into bond transparency, the COT-vs-TRACE comparison linked above is the natural next stop.