Trader Concentration in the COT Report: Who Holds the Position, Not Just How Big It Is

Key takeaways

• The COT Index answers how crowded a position is. It says nothing about who is holding it.
• The CFTC publishes the number of traders in each category and its own concentration figures, the share of open interest held by the four and eight largest traders. Almost no COT tool surfaces either.
• On the 1 September 2026 report, corn and wheat both read COT Index 100, maximum crowding. Corn's concentration sat in the 65th percentile of its own three-year record; wheat's sat in the 5th.
• Concentration must be read against a contract's own history. A top-four share that is routine in a thin softs market would be extraordinary in crude.
• This is descriptive. It tells you how a position is held, not which way price goes next.

Introduction

Two markets can print the same positioning extreme and be nothing alike underneath.

A COT Index of 100 tells you speculators are as long as they have been in three years. It does not tell you whether that length sits with 150 funds each holding a modest line, or with 20 holding the same total. Those two markets behave differently when the position is unwound, because the second one needs far fewer decisions to move.

The CFTC has published the data to tell them apart since the Disaggregated report began. It sits in the same file as the numbers everyone reads, in columns almost nobody opens.

What the CFTC Actually Publishes

The Disaggregated report carries 191 columns for commodity markets. Most COT tools read about twenty of them: the position columns for each trader category, plus open interest.

Two other families matter here.

Trader counts. For each category and each side, the CFTC reports how many traders hold a position. Managed Money long, Managed Money short, Producer/Merchant long, and so on. On the 25 August 2026 report, WTI crude showed 66 Managed Money traders long and 34 short.

Concentration. Separately, the CFTC publishes the percentage of open interest held by the four and by the eight largest traders, both gross and net. This is the exchange's own measurement of the book, not an estimate derived from category totals.

Neither is hidden. Both ship in the same weekly zip file as everything else.

Average Position Size, and Why It Is Not the Same as Net Position

Divide a category's gross position by its trader count and you get the average position per trader. Express that as a share of open interest and it becomes comparable across markets and across time, which raw contract counts are not, because open interest grows over the years.

The reading answers a question net position cannot: how much conviction, or leverage, sits behind each participant?

A market where the speculative cohort has grown while average position size has stayed flat is one where new money arrived. A market where the cohort shrank while the total position held is one where the remaining holders got bigger. The second is more fragile, not because the position is larger, but because it is held by fewer decision makers.

Watch the participation trend alongside it. Traders leaving while a position persists is a structural change that the level alone never shows.

Reading Concentration Against the Contract, Not Across Markets

This is the mistake to avoid.

Absolute concentration is not comparable between contracts. A top-four net share that is unremarkable in a thin softs market would be extraordinary in crude, because the underlying liquidity is different by an order of magnitude. Comparing the raw percentages across markets produces a ranking of market size, not of crowding.

The fix is to rank each contract against its own record. COTInsight uses a three-year window, the same window behind the COT Index, so "high concentration" means high for this contract, judged the same way "crowded" is.

The Corn and Wheat Example

From the report dated 1 September 2026, both markets at the same extreme:

Identical headline. Maximum speculative length in both. But wheat's crowding was spread across an unusually wide set of holders, sitting in the fifth percentile of its own concentration record, while corn's was middling for corn.

A dashboard that shows only the index renders those two markets identically. They are not identical.

The same week, ICE gasoil was crowded long at COT Index 85.2 with concentration in the 1.9th percentile of its own history: about as broadly held as that contract has ever been while carrying that much length.

What This Does Not Tell You

It does not forecast direction, and nothing here should be read as if it did.

Concentration describes how a position is held. Whether a concentrated position unwinds violently, drifts sideways, or simply keeps going depends on the catalyst, the curve, the fundamentals and the price level, none of which are in this data. A position held in few hands is not a sell signal, and a broadly held one is not a green light.

What it does is stop two very different markets from looking the same on your screen. That is a real gain, and it is a smaller claim than a forecast.

One more caution, and it is the reason this feature took care to build. The CFTC writes a full stop rather than a number when too few traders hold a position to disclose the count. That is not zero. NYMEX Brent Last Day carries that marker on the current report, alongside 318,384 open interest and a live Managed Money position. Treating a withheld count as zero would produce a confident, wrong statement about a market the exchange is simply declining to break down. Where the count is withheld, COTInsight shows nothing rather than a fabricated figure.

Where to Find It on COTInsight

Ultimate subscribers see positioning structure on every instrument detail view, in the Participant Interpretation panel:

• the spec trader count and which way it is trending
• average position size per trader as a share of open interest
• the CFTC's top-4 net concentration, with its percentile against that contract's own three-year history
• a plain-language reading crossing the positioning extreme with the concentration rank, from "Crowded long, few hands" to "Crowded long, broadly held"

All of it exports with the rest of the board in the Ultimate CSV and the PDF briefing, so the structure travels with the numbers rather than living only on screen.

The COTInsight TradingView indicator, also Ultimate, puts the z-score, COT Index, regime and momentum on your own weekly chart. The concentration layer stays on the dashboard, where the full participant breakdown sits beside it.

Frequently Asked Questions

Does the CFTC publish trader counts for every market?

For most, but not all. Where too few traders hold a position to disclose the number, the count is withheld and shown as a full stop in the raw file. Those markets get no structure reading rather than a zero.

Is concentration the same as the Herfindahl index some tools show?

No. A Herfindahl computed over the five participant categories measures how the pie splits between groups. The CFTC's concentration columns measure what share of open interest the largest individual traders hold. Different questions, and COTInsight reports both.

Why rank concentration rather than show the raw percentage?

Because the raw percentage is not comparable between markets. Ranking it against the contract's own three-year record makes "high" mean high for that contract, which is the only way the comparison is meaningful.

Does high concentration predict a reversal?

No, and COTInsight does not claim it does. It describes market structure. Timing needs the curve, the fundamentals and the price action alongside it.

Do other COT tools show trader counts and concentration?

The columns are in the same free CFTC file every tool downloads, and almost none surface them. Where each product stops is set out in best COT report tools compared.

Summary

The COT Index tells you how crowded a market is. Trader counts and concentration tell you who is doing the crowding. The CFTC publishes both every Friday, in the same file, and the second half is mostly ignored.

Corn and wheat both hit maximum speculative length on 1 September 2026. One was held in ordinary hands for that market, the other spread about as wide as wheat gets. On a dashboard showing only the index, those two weeks look the same.