COMEX · 100 troy ounces · ticker GC · CFTC code 088691 · Metals
Non-commercial longs minus shorts, 2023-11-14 to 2026-11-03. Speculators were net long throughout this window.
As of the 3 November 2026 report, large speculators were net long 179,548 contracts in Gold futures — 220,530 long against 40,982 short. That places the market at a COT Index of 26.3, meaning current net positioning sits 26.3% of the way up its own three-year range, where 0 is the most net-short speculators have been since 2024 and 100 is the most net-long.
At 26.3, positioning is neutral — 16.3 index points away from the crowded-short threshold. There is no crowding precondition here, which in this framework means there is nothing to fade. Most markets are neutral most of the time; that is the point of measuring extremes.
The 26-week index of 28.4 sits close to the three-year reading, so positioning looks consistent across both horizons — the current stance has been in place for a while rather than being a recent shift.
Week on week, speculators cut 6,903 contracts of net exposure, and over the past four reports the net position has moved by -30,787 contracts. The largest single-week swing in the last three years was +49,657 contracts, in the week to 5 March 2024.
Historically, Gold has spent 15 of the last 156 weekly reports crowded long and 0 crowded short — 10% of the time at one extreme or the other. Markets that reach extremes often are less informative when they do; markets that rarely get crowded are worth more attention when they finally are.
The three-year extremes bracket the current reading: the most net-long speculators have been was +315,390 contracts in the week to 24 September 2024, and the most net-short was +131,168 contracts in the week to 13 February 2024. Open interest across all participants currently stands at 399,169 contracts, with speculative longs and shorts together accounting for roughly 65.5% of it.
The world's benchmark gold futures contract, each representing 100 troy ounces of refined gold. The primary price discovery venue for the global gold market.
Miners hedging forward production, refiners, and bullion banks managing physical inventory and OTC client flow.
Speculative gold positioning responds to real interest rates, the dollar, central bank buying and episodic safe-haven demand. Extremes typically build after extended trending moves rather than around single events.
Gold has one of the longest and cleanest COT histories available, and its speculative positioning has a well-documented tendency to peak near price highs. It is the market where the crowding-at-exhaustion pattern is easiest to see historically.
The COT Index is a normalisation, not a raw position. It answers one question: where does the current net speculative position sit inside its own recent range? A reading of 26.3 means the current net position is 26.3% of the way between the most net-short speculators have been in three years (0) and the most net-long they have been (100).
Normalising matters because raw contract counts are not comparable across markets or across time. Gold carries 399,169 contracts of open interest today; a net position that would be an extreme in a thin market is unremarkable in a deep one. The index strips that out, which is why a reading of 95 in Gold and a reading of 95 in WTI crude oil mean the same thing about crowding even though the underlying contract counts differ by orders of magnitude.
Two thresholds define the classification used here. A market is crowded long at an index of 90 or above and crowded short at 10 or below. Everything in between is neutral. Those cut-offs are conventional rather than magic — at 90, speculators are in the top decile of their three-year net-long range, which is a reasonable working definition of "most of the people who were going to buy have bought".
The 26-week index alongside it is a recency check. When the three-year index is extreme but the 26-week index is not, the crowding is old and has partly unwound. When both are extreme, positioning is stretched on every horizon that matters.
The CFTC splits reportable positions into commercial and non-commercial. Only the non-commercial ("large speculator") side is used here. Commercial hedgers take positions for structural business reasons — miners hedging forward production, refiners, and bullion banks managing physical inventory and OTC client flow. They are not expressing a view on price direction in a way that crowd psychology applies to, so their positioning is not a sentiment signal. Speculators are, and they tend to be most heavily positioned near the end of a move rather than the start.
Crowding is a precondition, not a trade. Jason Shapiro's process requires two further confirmations before a crowded market becomes a candidate: a news-reaction failure, where news that should help the crowd fails to move price in their favour, and price-action confirmation on the chart. Positioning alone has no timing information in it — a market can stay crowded for months, and this page shows exactly that history for Gold above. The data is also lagged: the CFTC publishes on Fridays at about 3:30pm ET with positions as of the prior Tuesday, so what you are reading is between three and nine days old depending on when you read it.
Source. CFTC Commitments of Traders
Legacy Futures-Only report, via the public Socrata API
(dataset 6dca-aqww). No account or key required. Figures on this page were written into it
when the page was last generated, not fetched in your browser.
Calculation. Net position = non-commercial long minus non-commercial short. COT Index = (current net − 156-week minimum) ÷ (156-week maximum − 156-week minimum) × 100. The historical crowding counts roll that same 156-week index across each of the last 156 weekly reports, which needs roughly six years of history; where a contract has less, the page says so.
Not investment advice. Everything here is for research and education. Crowded positioning is a precondition in a discretionary framework, not a signal, and nothing on this page accounts for your circumstances, risk tolerance or time horizon. Past positioning says nothing certain about future price.