What Does the COT Report Mean for Gold Prices?

What the report is actually counting

The Commodity Futures Trading Commission (CFTC) requires every large trader in gold futures to report their positions weekly. It sorts them into two broad buckets that matter for this discussion. "Non-commercial" traders — hedge funds, commodity trading advisors, other speculators — trade gold to express a view or capture a return; they have no underlying business need for the metal. "Commercial" traders — miners, refiners, bullion banks — use futures to hedge physical exposure, such as a mining company locking in a price for gold it hasn't dug up yet. Because futures are a zero-sum contract, every long position has a short on the other side, so the two groups' net positions move as mirror images of each other.

The number everyone quotes is the speculators' "net non-commercial" position: their long contracts minus their short contracts. As of the report dated September 22, 2026, that figure stood at 225,853 contracts, against a net commercial position of -262,903 contracts (commercial traders, as usual, sit net short because they're mostly hedging mine output). Total open interest — the number of futures contracts still outstanding, long or short — was 412,800 contracts.

Why the raw number needs context

A net long of 225,853 contracts means nothing on its own; it needs to be sized against the market and against history. Dividing the speculative net by open interest gives 54.7%, meaning speculative net length equals more than half of all outstanding contracts. On its own that's just a ratio. The more useful step is placing today's reading against the past year and the past three years of reports: speculative net length currently sits at the 90.4th percentile on both horizons. In plain terms, over the last three years, speculators have been more net-long than this only about one time in ten.

Read the table left to right: the first three rows are raw counts, the "% of open interest" row scales that count to the size of the market, and the percentile rows scale it again against history so you can tell a genuinely stretched reading from a routine one. The four-week change matters too — speculators trimmed roughly 17,500 contracts of net length in the month before this report, so the position is already unwinding somewhat even while it remains historically large.

Crowding is not a forecast

It's tempting to read "90th percentile net long" as "gold is about to fall," but that mixes up two different things: direction and fragility. A crowded long position means a large share of the market is positioned the same way, which raises the odds of a sharp, mechanical move if those holders decide to exit together — margin calls, stop-losses, or a shift in the macro story can all trigger that kind of unwind. It does not tell you when that will happen, or whether new buyers might absorb the selling and keep the crowded trade going for months. Our own review of this data found no reliable edge in using positioning extremes to time entries or exits, which is exactly why we describe it as a measure of asymmetry — the risk of a violent move in one direction — rather than a directional signal.

History in this same series backs that caution up. Net speculative length in gold futures fell to just 52,081 contracts in late September 2022, an extreme low, and then rebounded over the following year. It has swung between roughly 70,000 and nearly 300,000 contracts multiple times since 2019, and periods of "extreme" positioning by percentile have preceded both further gains and sharp pullbacks depending on what else was happening in the macro backdrop at the time — real interest rates, the dollar, and the Federal Reserve's policy stance among them. As of this report, real 10-year Treasury yields stood at 2.83% in a rising three-month trend, a restrictive backdrop that has coincided with gold trading in a confirmed downtrend, roughly 8.4% below its 200-day moving average as of September 29, 2026. Positioning data and price trend can agree or disagree, and when they disagree it's a reason for more caution, not a rule for which one to believe.

The report is always older than it looks

The CFTC's own disclosure states the constraint plainly: figures are collected as of Tuesday's close and published the following Friday, so the data is already up to three days old the moment it's released. In practice the gap is usually worse. The report referenced throughout this article carries positions as of Tuesday, September 22, 2026; by the time a reader opens it on a typical Friday or over the following weekend, and certainly by September 29 when gold and silver futures had already moved on from that snapshot — silver futures alone were down 7.4% over the prior seven trading days — the underlying positioning may look nothing like the week-old numbers being reported. A CFTC holiday delays the release by a further day, stretching the lag closer to a full week. If you want to track exactly when each week's report lands relative to other macro releases, the MacroGold calendar lists CFTC and economic release dates together so the lag is visible rather than a surprise.

Takeaway and limits

The COT report tells you how concentrated speculative positioning is in gold futures right now, and a reading near the 90th percentile — as of September 22, 2026 — flags a market where a large share of participants are leaning the same way, which raises the chance of an abrupt unwind if sentiment shifts. It does not tell you when that unwind might happen, whether it happens at all before positioning grows even larger, or which direction gold moves in the meantime. It also only covers futures and options on one exchange, so it misses physical bullion, ETF flows, and over-the-counter derivatives entirely. Treat it as one gauge of market fragility to read alongside real rates, the dollar, and price trend, not as a standalone signal, and remember that whatever number you're looking at is already several days behind the market by the time you read it.