Who Holds the Position

A commodity price tells you what the marginal trade cleared at. It does not tell you who was on each side, or whether they were hedging a cargo or expressing a view. That second question matters, because a market where the longs are all funds is fragile in a way that a market where the longs are all flour millers is not. The attempt to answer it produces the most widely quoted, most widely misread dataset in commodities: the Commitments of Traders report.

The American original

The US report is the oldest and the template for everything else. Its ancestry runs to the grain-market reforms of the 1920s: the Grain Futures Administration published its first comprehensive report of hedging and speculation in 1924. That first report was annual. Monthly publication did not begin until June 30, 1962, moved to twice monthly in 1990 and fortnightly in 1992, and only became weekly in 2000. The dataset traders now refresh every Friday afternoon spent its first four decades appearing once a year.

Today the CFTC publishes each Friday at 3:30pm Eastern, reflecting positions as of the preceding Tuesday close. A market is covered only where 20 or more traders hold positions at or above the reporting level. The threshold mechanic is regularly described wrongly, so it is worth stating exactly: it is triggered per trader but then captures that trader entirely. If a firm holds a reportable position in any single expiry, its whole position in that commodity is reported, “regardless of size.”

Four reports, not one

What most people call “the COT” is four different publications with different category schemes.

The Legacy report is the old two-way split: Commercial, Non-Commercial and Nonreportable, with a spreading column inside Non-Commercial. The Disaggregated report, launched September 4, 2009 with history rebuilt to June 2006, breaks the same universe into Producer/Merchant/Processor/User, Swap Dealers, Managed Money and Other Reportables. Traders in Financial Futures, from 2010, covers financial contracts with Dealer/Intermediary, Asset Manager/Institutional, Leveraged Funds and Other Reportables. A supplemental report adds an Index Trader category for a short list of agricultural contracts.

One structural point is worth more than it usually gets. The Disaggregated report can be summed back to the Legacy two-way split. The financial-futures report cannot, because its four categories draw from both sides of the old commercial line. They are not the same operation viewed at different resolutions, and mixing them produces numbers that reconcile to nothing.

Why “commercial” never meant hedger

Here is the flaw at the centre of the whole exercise, and the regulator has conceded it. In the Legacy report a bank running a swaps book against long-only commodity index investors was classified as Commercial, because it was hedging its own exposure. So passive financial money, the most purely speculative capital in the market by any ordinary use of the word, arrived in the data wearing a hedger’s coat. In index-heavy agricultural markets that was not a rounding error; it was enough to invert the apparent balance between commerce and speculation.

The Disaggregated report exists specifically to fix this. It implements a recommendation from the CFTC’s September 2008 staff report on swap dealers and index traders, work done in the aftermath of the crude oil spike of that year, to break swap dealers out of the commercial category. Anyone quoting pre-2009 “commercial” positioning as a hedging series is quoting something that does not mean what they think it means.

Two further cautions sit in the CFTC’s own notes. Classification is self-declaredon a form and reviewed rather than audited. And it is not even stable across markets: in the Commission’s words, “a trader may be classified as a commercial trader in some commodities and as a non-commercial trader in other commodities.” The same legal entity can sit on both sides of the speculator line at the same moment, in different pits.

Which is why the practitioner shorthand narrowed. When a market note says “the specs are long,” it now almost always means the Managed Money column of the Disaggregated report, not the old non-commercial line.

Europe measures a different thing

Europe built its own regime under MiFID II, which from 3 January 2018 required venues trading commodity derivatives to publish a weekly position report by category of holder. The five categories are set in law: investment firms or credit institutions; investment funds (a UCITS or an alternative investment fund manager); other financial institutions, including insurers and pension schemes; commercial undertakings; and, for emissions products only, operators with compliance obligations. Each splits long and short, and within that splits positions that are risk-reducing and directly related to commercial activity from everything else. Options count on a delta-equivalent basis.

The crucial difference is not the number of buckets. It is that Europe classifies by what kind of firm you are, and America classifies by what you are doing in that market. A physical trading house and a macro hedge fund can both be “investment firms” in Europe while sitting in completely different CFTC categories. The two datasets are not translations of each other, and adding them together silently mixes two definitions of the same word.

Europe also publishes far less than the rule appears to promise, because the threshold has two limbs: a contract needs 20 open position holders and gross open interest exceeding four times deliverable supply. Thin contracts produce nothing at all, and where fewer than five holders sit in a category, the count is suppressed. There is now divergence to track as well: the UK went its own way through the FCA’s reform package, with the main rules in force from 6 July 2026, narrowing position limits to a short list of critical contracts and moving limit-setting from the regulator to the venue, while keeping the weekly report broadly intact.

The rest of the world is a patchwork

The assumption that every major exchange publishes something COT-like does not survive contact with the evidence.

Japan is the quiet standout. JPX publishes a weekly open-interest report broken down by investor category that exists only for its commodity futures, in English, free, with continuous history since July 2020, covering gold, platinum, rubber, corn, soybeans, the oil products, Dubai crude, JKM LNG and a long list of electricity contracts. It has no reporting threshold at all, so it is a complete census rather than a sample of large traders. On that specific measure it is more comprehensive than either the American or the European report.

Brazilmoved backwards. B3’s open-interest-by-participant-type view was discontinued in April 2025, and the replacement attaches participant type only to volume, aggregated across the exchange rather than per contract. You can no longer ask whether funds are long B3 arabica. Korea publishes flows, not positions, which is a different thing wearing similar clothes. Hong Kong’s investor survey has not appeared for years. Malaysia publishes a monthly demography of who trades palm oil futures, but by residence and firm type, so a refiner hedging and a macro fund speculating land in the same bucket.

China is the biggest question and deserves a careful answer rather than a confident one. Chinese exchanges publish daily trading rankings by member firm, a genuinely useful disclosure that the American and European reports do not offer. But it answers a different question: it tells you which broker carries the position, not what kind of tradersits behind it, and one clearing member’s book can hold a refinery’s hedge and a fund’s punt in the same line. Whether any Chinese venue publishes a hedger-against-speculator split in the Western sense is not something this book can state either way; the exchange sites defeat automated checking, and the honest position is that we do not know rather than that it does not exist.

That last distinction matters generally. There is a real difference between confirmed absent, as with Brazil’s discontinued report, and not found. Much of the writing on global positioning data quietly converts the second into the first.

Stitching venues together

Because the same commodity trades in more than one jurisdiction, anyone wanting a global picture has to combine reports, and crude oil shows why this is forced rather than optional. ICE’s WTI contract does appear in the CFTC report. Brent does not.So a complete view of crude positioning cannot be assembled from the American data alone, no matter how carefully it is read; the Brent leg has to come from ICE’s own publication.

There is a trap waiting there for the careless. The CFTC report does contain a contract called Brent Last Day, a cash-settled New York look-alike. Anyone summing everything with “Brent” in the name picks up the wrong instrument and believes they have covered the market.

The hazards of stitching go beyond naming. American data is as of Tuesday; European and LME data is as of Friday. Any transatlantic aggregate therefore blends two snapshots taken three days apart, across a week that may contain the move you are trying to explain. Futures-only and futures-and-options series are separate publications with different histories, so two legs of a sum can silently sit on different bases. And the categories, as above, do not mean the same things.

The map is not the market

The London Metal Exchange publishes its own Commitments of Traders Report each Tuesday, reflecting the previous Friday’s positions, using four of the five MiFID II categories, since December 2017. Its own policy document contains the most candid sentence in this entire subject. The exchange notes that the report is materially affected by the classification of large users whose activities arguably fall into several categories, and that because members write OTC contracts with clients and net that business before placing it on-exchange, the published data will inevitably reflect “only a subset of the total activity conducted within the LME ecosystem.”

That is an exchange stating in its own rulebook that its position report does not see the whole market. It is not a scandal, it is the nature of the instrument. Positions move OTC, between affiliates, and across venues that report on different days into different categories, and no weekly snapshot assembled from self-declared classifications is going to capture all of it.

None of which makes the reports useless. Crowded positioning genuinely does precede violent unwinds, and knowing that managed money is at a record short in a market with a tightening physical balance is worth knowing. But the reports are a coincident description, not a leading signal, and they are published with a lag that rules them out as a timing tool even if they were. The academic literature testing COT positioning as a contrarian indicator has generally struggled to find exploitable predictive power.

The architecture is also not settled. The CFTC opened a formal review of the Commitments of Traders programme in May 2026, seeking comment on trader classification, the category definitions themselves and publication frequency. The most quoted dataset in commodities is, as of this writing, under active reconsideration by the agency that publishes it.

Read the reports for what they are. They are four decades of annual snapshots that became a weekly habit, built on categories designed for a grain market in the 1920s, patched once in 2009 when financial money had made the old ones misleading, imitated abroad using different definitions on a different day of the week, and absent entirely from some of the largest venues on earth. Within those limits they are the best answer anyone publishes to the question of who is on the other side.