The Commitment of Traders (COT) report is a weekly disclosure published by the U.S. Commodity Futures Trading Commission (CFTC) that breaks down the open interest of major U.S. futures markets by trader type — splitting positions held by commercial hedgers, large speculators, and (in the newer reports) swap dealers, managed money, and asset managers. Its appeal as a sentiment tool rests on a simple premise: if you know who is positioned which way — the commercial "smart money" that handles the physical commodity versus the trend-following speculators — you can read crowd extremes. The core tension is that this premise, intuitive as it is, has held up far better in trading folklore than in rigorous testing.

How it's calculated / formed

The COT is not a model or formula; it is mandatory reporting. Any trader whose position in a market exceeds a CFTC reporting level must report it, and the CFTC aggregates those positions by category. A market appears in the report only if it has 20 or more reportable traders (CFTC). There are four report families:

• Legacy (data back to 1986): two buckets — commercial (hedgers/physical-market participants) and non-commercial (large speculators), plus a non-reportable residual for small traders.
• Disaggregated (physical commodities, data from June 2006): splits the old commercial bucket into Producer/Merchant/Processor/User and Swap Dealers, and the old non-commercial bucket into Managed Money and Other Reportables.
• Traders in Financial Futures (TFF) (financials — equity indices, rates, currencies, VIX): Dealer/Intermediary, Asset Manager/Institutional, Leveraged Funds, Other Reportables.
• Supplemental/CIT: adds an Index Trader category for 13 agricultural contracts.

Each category is reported as gross long, gross short, and spreading. The data snapshot is Tuesday's close of business, but the report is released Friday at 3:30 p.m. ET — a three-day lag baked into the data (CFTC). Most analytical use derives a net position (long minus short) per category and tracks its change week to week.

How it's used in practice

Raw net positions are hard to compare across markets, so practitioners normalize them with the COT Index (popularized by Larry Williams and Stephen Briese): ((current net − lowest net) / (highest net − lowest net)) × 100 over a lookback window, scaling positioning to a 0–100 percentile. A reading near 0 means a group is the most short/bearish it has been in the window; near 100, the most long/bullish. Williams commonly applies a ~3-year lookback with 25/75 thresholds; Briese's histogram variant uses 5/90 (per The Commitments of Traders Bible).

The dominant interpretive frame is contrarian on speculators, confirmatory on commercials. Commercials are treated as informed because they live in the physical market and tend to lean against price trends (buying weakness, selling strength); large speculators are treated as a momentum crowd that piles in at extremes. The classic setup: commercial net long at a multi-year high while large specs are net short at an extreme — read as a possible bottom. The mirror image flags possible tops. Analysts watch for extremes and divergences, not absolute levels, and usually require price confirmation rather than trading the report in isolation. In equity-index and currency contexts the same logic is applied to TFF leveraged funds (the speculative crowd) and asset managers.

Adoption, debate & evidence

COT is one of the most widely watched sentiment datasets among futures and FX traders, and the "follow the commercials, fade the speculators" heuristic is near-canonical in that community. But here the honest distinction matters: the folklore is far stronger than the measured edge.

The most rigorous skeptical work is Sanders, Irwin & Merrin, "Smart Money: The Forecasting Ability of CFTC Large Traders in Agricultural Futures Markets" (Journal of Agricultural and Resource Economics, 2009). Using bivariate Granger-causality tests across 10 agricultural markets, they found "very little evidence that traders' positions are useful in forecasting returns," while there was substantial evidence that traders respond to prices — i.e., returns lead positions, not the reverse, with non-commercials in particular behaving as trend-followers. Their broader conclusion: the results "generally do not support use of the COT data in predicting price movements." Related CFTC-data studies reach the same causal verdict — positioning is largely a reaction to price.

This directly undercuts the most common practitioner claims. Widely circulated figures like "60–70% win rates following commercials at extremes" appear in trading literature and vendor marketing but typically lack peer-reviewed support, transparent out-of-sample testing, or transaction-cost accounting; they should be treated as unverified folklore, not established base rates. There is some academic nuance on the other side — e.g., work in the cross-listing/equity literature suggests commodity-futures positioning information can be priced into related stocks with delay — but evidence that retail-accessible COT signals generate a tradable edge net of costs is weak and not robust across markets or time.

Strengths & limitations

Genuine strengths: COT is actual position data, not survey opinion — it shows where real money is committed, and it is free, standardized, and decades-deep, making it excellent for context (is positioning crowded or washed out?) and for understanding market composition. Extremes do mark crowded conditions that often precede reversals, even if the timing is unreliable.

Limitations are serious. (1) The three-day lag plus weekly frequency means you never see live positioning — by Friday the Tuesday picture may be stale. (2) Causality runs the wrong way for naive forecasting: positions follow prices. (3) Category labels are imperfect — swap dealers and index traders can sit in the "commercial" bucket despite being financial, blurring the smart-money story. (4) Spread positions and the lack of price/leverage detail hide a lot. The #1 misuse is treating an extreme COT Index reading as a standalone timing signal — extremes can persist for months in a strong trend, and fading them mechanically is a documented way to get run over.

Sources

• CFTC, Commitments of Traders (report types, categories, 20-trader threshold, Tuesday snapshot / Friday 3:30 p.m. release): https://www.cftc.gov/MarketReports/CommitmentsofTraders/index.htm
• CFTC, Disaggregated Explanatory Notes (Producer/Merchant/Processor/User, Swap Dealers, Managed Money, Other Reportables): https://www.cftc.gov/MarketReports/CommitmentsofTraders/DisaggregatedExplanatoryNotes/index.htm
• CFTC, Traders in Financial Futures Explanatory Notes (TFF categories): https://www.cftc.gov/sites/default/files/idc/groups/public/@commitmentsoftraders/documents/file/tfmexplanatorynotes.pdf
• Sanders, Irwin & Merrin (2009), "Smart Money: The Forecasting Ability of CFTC Large Traders in Agricultural Futures Markets," J. of Agricultural and Resource Economics — key skeptical finding (Granger causality; positions react to prices): https://ageconsearch.umn.edu/record/54547/
• "Does the CFTC Commitments of Traders Report Contain Useful Information?" (returns lead positions; lack of predictive power): https://ideas.repec.org/p/ags/ncrtci/18929.html
• Stephen Briese, The Commitments of Traders Bible — COT Index methodology and contrarian thresholds: https://onlinelibrary.wiley.com/doi/book/10.1002/9781119201861

Dispute flag: Practitioner literature (Williams, Briese, vendor sites) frames COT as a high-win-rate contrarian edge; peer-reviewed evidence (Sanders/Irwin et al.) finds little forecasting power and reverse causality. The popular win-rate figures are unverified. This doc sides with the academic record on predictive claims while acknowledging COT's value for context/positioning.