Video · 10:50 · 16 November 2026

COT Report for Gold (XAUUSD): How to Read the Commitments of Traders

How to read the Commitments of Traders (COT) report for gold (XAUUSD), from zero to the pro level, on real CFTC data. You will learn what the COT report is and who publishes it (the CFTC, the US futures regulator), the five trader groups in the disaggregated report (producers and merchants, swap dealers, managed money, other reportables, non-reportables), why the data is already 3 days old when it comes out (positions as of Tuesday, released Friday 3:30 p.m. New York time), how to read net positions, longs and shorts on their own, and open interest, the price and open interest 2x2 grid, what a crowded trade is and how to measure it with a 5-year percentile, why producers and swap dealers are almost always short (hedging), a real divergence between COT and price, what COT cannot do, an honest test of a simple contrarian rule (it failed), and how a pro uses COT in a weekly routine: as context for size and patience, never as an entry.

Chapters

  • 0:00A record fund position and what gold did next
  • 0:54What the COT report is and the five trader groups
  • 1:57When it comes out: Tuesday positions, Friday release
  • 2:30Net positions, longs and shorts, and open interest
  • 3:46The crowded trade: 5-year percentile, measured
  • 4:58Why producers and swap dealers are usually short
  • 5:46When COT and price disagree (a real divergence)
  • 6:42What COT cannot do
  • 7:37An honest test: fading the crowd
  • 8:49How a pro uses COT: the weekly routine
  • 9:30How should COT change my trade? (flowchart)
  • 10:03Recap

Key points

  • The COT report from the CFTC shows who holds COMEX gold futures: producers and merchants, swap dealers, managed money (the funds), other reportables and small traders.

  • It shows positions as of Tuesday and comes out Friday at 3:30 p.m. New York time, so it is already 3 days old.

  • A 5-year percentile shows when the funds are crowded long or short; after crowded readings gold did about the same as after any week, so there was no clear edge.

  • Producers and swap dealers are short almost every week because they hedge, not because they bet on a fall.

  • Our honest test of fading the crowd failed (24 trades, -12.9% in total). Use COT as weekly context for size and patience, never as an entry signal, and never as a promise of profit.

Chart data: CFTC Commitments of Traders, disaggregated futures-only report, COMEX gold (088691), weekly from June 2006 to 3 November 2026; COMEX gold futures daily closes from Yahoo Finance (GC=F). Education only, not a signal.

Read more: Cot positioning · Gold pro track · Cot report · Managed money · Open interest · Swap dealers · Commercials

Transcript

Tuesday, 9 October 2018. The big funds in gold futures held their biggest bet against gold since this report began in 2006: net short 109,454 contracts. Gold closed at 1,187.2. By the time the report came out, gold had already risen 39.2 points. And a year later, it was up 21.6%. The funds' biggest bet on gold came in July 2016. A year later, gold was 10.5% lower. Two records, two perfect calls for doing the opposite. So should you just fade the funds? We tested it on fifteen years of data, and the answer is no. In this lesson: what the COT report is, who's in it, how to read it, crowded trades, what it can't do, an honest test, and how a pro really uses it.

Chapter one. What the COT report is, and who's in it. The CFTC is the US regulator for futures markets. Every week, it counts who holds COMEX gold futures, and publishes the totals. This is the Commitments of Traders report, or COT. We use the disaggregated version, futures only. It splits traders into five groups. There's also an older legacy version, with just commercials and non-commercials. Never mix numbers from the two. Producers and merchants: miners, refiners and bullion dealers. Swap dealers: mostly banks. Managed money: hedge funds and trading advisers. Other reportables: other large traders. And non-reportables: everyone too small to report. Here's the latest report in our data, for 3 November 2026. Managed money held 122,283 long contracts and 18,669 short. Swap dealers were short 215,291.

Chapter two. When it comes out. The positions are counted on Tuesday. The report comes out on Friday, at 3:30 in the afternoon, New York time. So when you read it, it's already 3 days old. You use it until the next Friday, so by then it's up to 10 days old. Holidays can delay it. Remember 2018: between that Tuesday and the first close after the report, gold had already moved 39.2 points. So every test in this lesson starts at the first close after the release, never at the Tuesday price.

Chapter three. Net positions and open interest. Net means longs minus shorts. For managed money, 122,283 minus 18,669 is a net long of 103,614 contracts. One contract is 100 ounces, so that's about 322 tonnes of gold. Always check both sides. That week the net fell 1,948. Longs fell only 225, but shorts rose 1,723. So new sellers did most of it. Open interest is the number of contracts still open. It rises when new positions open, and falls when they close. On 3 November, it was 399,169, up 2,359. Now put price and open interest together. Price up, open interest up: new buying. Price up, open interest down: short covering. Price down, open interest up: new selling. Both down: longs giving up. That week, gold fell 134.2 points while open interest rose. New selling. And the managed money shorts agree.

Chapter four. The crowded trade. A trade is crowded when a group is far more one-sided than usual. To measure usual, compare today's net with the last five years: 260 weekly reports. That's the percentile. In the top 10%, funds are crowded long. In the bottom 10%, crowded short. On 3 November 2026, managed money sat at the 47th percentile. Right in the middle. The idea: when everyone is already long, who's left to buy? So a crowded long should be a warning. Let's measure it. From 2011 to 2026, there were 73 crowded long weeks. 13 weeks later, gold was up 1.4% on average, and higher in 39 of them. After 123 crowded short weeks, gold was up 1.5%. And after any week at all: 1.9%. So after crowded longs, gold still rose, just a little less than usual. After crowded shorts, about the same as any week. And these weeks come in runs: only 15 separate crowded long spells, and 23 short ones. That's a small sample.

Chapter five. Why producers are usually short. A gold miner digs gold it will sell next year. To lock in today's price, it sells futures now. That's a hedge, not a bet that gold will fall. Swap dealers are mostly banks. When clients buy gold through the bank, the bank sells futures to cover itself. So its short mostly mirrors what clients want, not the bank's own view. So in 1,065 weekly reports since 2006, producers were net short in 1,056 weeks, and swap dealers in 1,007. Their weekly changes mirror managed money almost exactly: a correlation of minus 0.92. When funds buy, the hedgers are on the other side. Their short is normal, not a bearish signal.

Chapter six. When COT and price disagree. 24 September 2024. Managed money was net long 219,029 contracts, the 97th percentile. A crowded long, with gold at 2,677. Then the funds cut back, but gold kept climbing. On 10 March 2026 it closed at 5,242.1, up 95.8%, while the funds held only 98,399, the 50th percentile. Price made a much higher high. Positioning made a much lower one. That's a divergence: other buyers drove this rally, not the funds. It did end. By 3 November 2026, gold was 1,301.7 points lower. But the warning came 18 months early. Anyone who sold on it was run over first.

Chapter seven. What COT cannot do. It's not a timing tool. Extremes can last for months. In 2013, managed money sat in the crowded short zone for 25 weeks in a row. And it was still net long: 76,501 contracts on 29 January 2013. Crowded only means far from its own last five years. Over those weeks, gold fell from 1,660.7 to 1,243.6. That's 417.1 points, or 25.1%. It's late: 3 days old on release. And it covers COMEX futures only. London trading, ETFs and most options aren't in it. A fund can own gold through an ETF and sell futures as a hedge: COT only sees the short. It shows that positions changed, never why.

Chapter eight. An honest test. The rule: when managed money first enters the bottom 10%, buy. When it first enters the top 10%, sell. Enter at the first close after the release, hold 13 weeks, one trade at a time, with costs of 0.30 per ounce. From 2011 to 2025, that made 24 trades. The 14 buys made 31.3% in total, about 2.2% a trade. 7 made money and 7 didn't. That's about what simply holding gold for any 13 weeks made. The 10 sells lost 44.2% in total, and only 2 made money. Selling crowded longs in 2024 and 2025 lost 12.6% twice, as gold kept rising. Overall, minus 12.9%. A 4 week hold lost too. And the famous 2018 record? The rule never got it. It had already bought in May 2018, and lost 7.1%. So the simple contrarian rule failed. Crowded positioning did not tell us when to buy or sell gold. And gold rose a lot over these years, which helped every buy and hurt every sell.

Chapter nine. How a pro uses COT. Once a week, after the Friday release, take ten minutes. Note the Tuesday date. Then managed money's net and its change. Did longs or shorts move? Check the 5-year percentile, and the price and open interest box. Check that the hedgers are still the usual mirror. Then write one line in your journal. For example: managed money neutral, 47th percentile, new selling, no crowding. Use it as context, not as an entry. If the crowd is on your side at an extreme, trade smaller, take profits sooner and don't add. Your entry still comes from your chart.

Chapter ten. How should COT change my trade? Is managed money in its top or bottom 10% of five years? If not, COT is neutral. Trade your normal plan. If it is, does your trade go with the crowd? Then cut the size, take profits sooner, and don't add. Is your trade against the crowd? Then wait until your chart confirms the turn. A crowded reading alone is not a reason to trade. Either way: normal setup, normal stop, and write the COT line in your journal.

Let's recap. The COT report shows who holds COMEX gold futures, as of Tuesday, released on Friday. It's 3 days old when you get it. Read the net, the longs and shorts on their own, and open interest. Producers and swap dealers are short because they hedge, and that's normal. Crowded positioning is context. Extremes can last for months, and our simple contrarian test lost. Use COT for size and patience, not to time entries. The full written guide is on xauusdglobal.com. Subscribe for the next lesson, and come trade with us in the XAUUSD Global community. It's free during early access.