Commitment of Traders Report: Why It's Already Too Late

Every Friday afternoon, someone posts a screenshot of the Commitment of Traders report claiming large speculators just hit a historic extreme in gold, and that a reversal is close. You pull up the chart. The move they're describing already

Every Friday afternoon, someone posts a screenshot of the Commitment of Traders report claiming large speculators just hit a historic extreme in gold, and that a reversal is close. You pull up the chart. The move they're describing already happened three days ago.

That's the trap built into the Commitment of Traders report (COT), the CFTC's weekly count of who holds what in US futures markets. It's real, government-filed data on actual positions, not another indicator built from price like everything else on your chart. But the report you read on Friday describes Tuesday's market, and a peer-reviewed backtest of trading purely off it found the edge mostly disappears once you try to run it as a system instead of a story.

The report is a photograph of last Tuesday, developed by Friday

The Commodity Futures Trading Commission (CFTC, the US regulator for futures and options markets) releases the COT report every Friday at 3:30 p.m. Eastern time. The positions inside it are frozen as of the previous Tuesday's close, per the CFTC's own release schedule.

Do the math and the data is roughly three business days old the moment you're allowed to see it. Whatever those same traders did Wednesday, Thursday, or Friday morning, the exact days you're most likely reacting to news in, isn't in the report you're staring at.

It's the equivalent of getting handed Tuesday's poker table talk on Friday night and being told to bet on tonight's hand. The read might still be useful context. It is never a live signal, because live was three days ago.

Fed, NFP, and CPI prints have the same structural problem: by the time the number everyone's watching gets published, the market that will react to it has already started pricing in the leak, the guess, and the positioning around the guess. COT just turns that lag into a fixed, published fact instead of a fuzzy one.

Three counting systems, not one "smart money" bucket

Retail traders who mention COT usually talk about it like there's one "smart money vs. dumb money" split. There isn't. The CFTC runs three separate formats, and which one applies depends on what you're trading.

The legacy format, run for every market, splits everyone into three buckets: commercial (hedgers using the market to manage real business risk), non-commercial (large speculators, mostly funds), and nonreportable (everyone below the CFTC's size threshold, retail traders included). It's the oldest and simplest cut.

Physical commodities, gold, oil, corn, natural gas, get the disaggregated format instead, in place since September 2009. It breaks the same open interest into four groups: Producer/Merchant/Processor/User, Swap Dealers, Managed Money, and Other Reportables.

Currencies, stock indices, and interest rate futures use a third format entirely: Traders in Financial Futures (TFF). Its four large-trader categories are Dealer/Intermediary, Asset Manager/Institutional, Leveraged Funds, and Other Reportables, on top of a Nonreportable bucket for everyone smaller.

None of the four financial categories is labeled "smart money." "Managed Money" specifically means a registered commodity trading advisor, commodity pool operator, or an unregistered fund the CFTC has identified, a defined legal category, not a vibe. That's worth sitting with if you've spent time around claims of invisible institutional footprints on a chart that never name who's supposedly leaving them. COT actually names the trader types. It just names them three days late.

A market only gets covered at all once 20 or more traders hold positions at or above the CFTC's reporting level. Drop below that count and the market quietly stops appearing until it qualifies again. Thinly traded futures contracts can vanish from the report entirely for stretches at a time.

The report never sees spot forex or CFDs

Here's the part that trips up the most people reading this on a phone with a CFD or spot forex account open. The COT report only counts positions in exchange-listed futures contracts, the ones cleared through CME and similar exchanges. It has no visibility into the interbank spot market or the CFD books retail brokers run internally.

We've written before about how different a CFD actually is from the instrument underneath it, and that's exactly why it matters here. EUR/USD futures positioning on CME correlates loosely with sentiment in spot EUR/USD, because the same macro forces push both. Loosely is not the same as directly. A "historic extreme" in currency futures is a real number about a real market, just not the specific market your account is actually exposed to.

A peer-reviewed backtest found the edge mostly disappears at scale

This is the part that separates COT from a Discord screenshot: someone actually tested it properly. A 2023 study published in the International Journal of Financial Markets and Derivatives built short-term reversal strategies using nothing but COT positioning data across a range of US futures markets.

Run long-only, the strategy produced statistically significant results in six individual markets. Extended to long-and-short, two markets beat a buy-and-hold benchmark by a significant margin. That sounds like a system worth running.

Then the authors combined those markets into a single portfolio, the step that turns "some interesting backtests" into "a strategy," and measured it against a plain S&P 500 buy-and-hold approach with a Sharpe ratio of 1.07. The portfolio underperformed. The authors' conclusion is that the CFTC's report contributes to more efficient derivatives markets, and that the strategy parameters could not generate excess Sharpe ratios (return per unit of risk) in a portfolio.

That result lines up with something we keep hammering on in our myths piece: an indicator, and COT positioning functions as one here, describes what already happened. A handful of markets showing significance in isolation is exactly what you'd expect from testing enough markets, not proof of a mechanism you can bank on.

Extreme positioning stories work backward, not forward

The other common pitch is the divergence story: large speculators are at a record long, commercials are heavily short against them, and "something has to give." Sometimes it does. The story only gets told about the times it worked.

The lag makes this worse, not better. By the time Friday's report shows you Tuesday's extreme, price has had three more sessions to either extend the move or start unwinding it. You're not looking at a warning. You're looking at a photograph of a moment that already resolved one way or the other by the time you saw it.

The Commitment of Traders report isn't lying to you. It's just telling you about a market that moved on three days ago and never waited for you to catch up.

— The Tradoki desk note

What the report is actually good for

None of this makes COT worthless. It makes it a context tool, not an entry trigger. Positioning extremes are a useful input for a weeks-long thesis about which way a market is leaning structurally, not a trade signal by itself.

Used that way, COT sits next to your own read of price and whatever risk you've already sized the position for, not in place of either one. It can tell you the crowd is leaning hard one direction. It can't tell you when, or whether, that lean breaks, and it definitely can't tell you that before the break has already started.

● FAQ

What is the Commitment of Traders (COT) report?

It's a weekly report the CFTC publishes on US futures markets, breaking down open interest (the total live contracts nobody has closed yet) by trader type: commercial hedgers, large speculators, and everyone below the reporting threshold. It's built from actual position filings, not from price or volume, which is what separates it from a normal chart indicator.

How old is the data in the COT report by the time you read it?

About three business days. The report published Friday at 3:30pm Eastern reflects positions as they stood at Tuesday's close, per the CFTC's own release schedule. Whatever those traders did Wednesday through Friday morning isn't in it yet.

Does the COT report predict when a trend will reverse?

Not reliably enough to trade on its own. A 2023 peer-reviewed study built short-term reversal strategies purely from COT data across a range of US futures markets, found statistically significant results in a handful of individual markets, but the same approach run as one portfolio underperformed a simple S&P 500 buy-and-hold benchmark. Isolated wins don't automatically survive being generalized into a system.

Does the COT report cover spot forex or CFD trading?

No. It only covers positions in exchange-listed futures contracts, like CME currency futures, not the interbank spot market or the CFD books retail brokers run. The two markets move together loosely, but a futures positioning extreme isn't a direct readout of what's happening in the CFD account you're actually trading.

Can a market disappear from the COT report?

Yes. The CFTC only publishes a market once 20 or more traders hold positions at or above its reporting threshold. If a contract's reportable large-trader count drops below that, the report stops covering it until it qualifies again.