Reviewed by Isaac Kahan, Owner ·
By Isaac Kahan, Owner, Bullion Trading LLC
When hedge funds were most bearish on gold, gold usually rose over the following year. When they were most bullish, it gave no reliable warning. Bullion Trading LLC research matched the gold COT report, the CFTC’s weekly Commitments of Traders data, with World Bank monthly gold prices for the 244 months from June 2006 to September 2026. In the 24 months when hedge funds held their smallest bets on gold, or bets against it, gold’s monthly average was higher a year later 22 times (92%), against 68% for all months. In the 24 months when their bets were biggest, it was higher 15 times (63%).
That sounds like a trading rule, but it is weaker than it looks. The 24 bearish months fall into only 10 separate episodes, the edge shrinks on what a trader could have known at the time, and in the 1990s the speculators betting against gold were right for years.
| Key finding | Result |
|---|---|
| Data | Managed money (the CFTC’s hedge fund group) in COMEX gold futures: 1,064 weekly reports, June 13, 2006 to October 27, 2026 |
| Gold higher a year after the most bearish 10% of months | 22 of 24 (92%), median gain 15.1% |
| After the most bullish 10% | 15 of 24 (63%), median gain 8.2% |
| After any month | 157 of 231 (68%), median gain 7.9% |
| Separate episodes higher a year later | Bearish: 8 of 10. Bullish: 4 of 9 |
| Calendar-shift chance test | 4% of shifts matched the bearish 92%, 37% its median gain |
| Only data known at the time (from May 2009) | Bearish months: 24 of 31 higher (77%), against 65% |
| Broader speculator data, 1986 to 2026 | Bearish months, all 1988 to 2001: 21 of 49 higher (43%), against 61% |
| Latest report (November 3, 2026) | Net long 103,614 contracts, 26.0% of open interest: not an extreme |
What the gold COT report shows about hedge funds
The gold COT report counts who holds COMEX gold futures and on which side. The CFTC usually publishes it on Fridays at 3:30 pm Eastern Time, with data from the preceding Tuesday. Since September 4, 2009, its disaggregated report has split large traders into four groups, one of them managed money: “a registered commodity trading advisor (CTA); a registered commodity pool operator (CPO); or an unregistered fund identified by CFTC.” Traders call it hedge funds.
Traders watch the net position, longs minus shorts. Each COMEX gold contract covers 100 troy ounces, so the biggest net long in this data, 273,076 contracts on July 5, 2016, was a bet on about 850 tonnes of gold. As the market’s size changes, we measure it as a share of open interest (all contracts outstanding): 652,971 that week and 399,169 on November 3, 2026.
When hedge funds were most bearish, gold usually rose
The most bearish months are the 24 with hedge funds’ net long at 6.46% of open interest or less, including every month they ended net short. Fifteen fell in 2015, 2018 and 2022. Managed money first turned net short on July 14, 2015, by 1,123 contracts; news reports dated it to the week to July 21 (by CFTC data, a net short of 13,556). Its biggest net short came on October 9, 2018: 109,454 contracts, 23.5% of open interest. In August 2018, as the broader speculator group turned net short, Peter Boockvar of Bleakley Financial Group told CNBC: “It’s tough to find a more contrarian indicator.”
He was right that time, and usually: gold’s monthly average was higher six months after 23 of the 24 bearish months, against 67% for all months, and a year after 22. Eight of the 10 separate bearish episodes were followed by gains. The misses were late 2013, when gold rallied into March 2014 but ended the year 1.7% lower, and September 2021 (down 6.3%).
When hedge funds were most bullish, there was no reliable warning
The common view is that crowded longs warn of a pullback. The Perth Mint wrote in 2020 that managed money traders “often get overextended at or near market inflection points,” so a big build-up of longs could signal “a short-term price correction.” In this data, after the 24 most bullish months (net longs of 36.45% of open interest or more), gold was higher three months later only 54% of the time, against 62% for all months.
The summer of 2016, hedge funds’ biggest net long in contracts, preceded a 7.5% fall over the next year. But crowded longs in 2009, 2010 and 2024 came early in big rallies: gold was 28.1%, 25.5% and 42.7% higher a year after each. Four of the nine bullish episodes were followed by gains, including the largest of all 19.
Every hedge fund extreme in gold since 2006
All 19 episodes, merging extreme months less than three months apart:
| Extreme months | Side | Peak report | Net contracts | % of open interest | Gold 6 months later | Gold 12 months later |
|---|---|---|---|---|---|---|
| Oct 2006 | Bearish | Oct 31 | 20,892 | 6.5% | +6.3% | +28.4% |
| Feb 2008 | Bullish | Feb 26 | 186,305 | 37.6% | -14.3% | -4.6% |
| Jul to Dec 2009 | Bullish | Sep 29 | 194,449 | 42.8% | +15.4% | +28.1% |
| Apr 2010 | Bullish | Apr 27 | 205,443 | 38.1% | +13.6% | +25.5% |
| Mar to Jul 2011 | Bullish | Jul 26 | 213,501 | 39.8% | +12.5% | +11.4% |
| Feb 2012 | Bullish | Feb 28 | 177,143 | 37.0% | +4.1% | -4.9% |
| Nov to Dec 2013 | Bearish | Nov 26 | 15,961 | 4.0% | +4.7% | -1.7% |
| Jun to Jul 2015 | Bearish | Jul 28 | -14,633 | -3.3% | -2.7% | +18.5% |
| Nov 2015 to Jan 2016 | Bearish | Dec 29 | -27,201 | -6.8% | +18.7% | +7.6% |
| Jun to Sep 2016 | Bullish | Sep 27 | 255,177 | 43.8% | -10.8% | -7.5% |
| Aug 2017 | Bullish | Aug 29 | 221,126 | 41.0% | +0.8% | -8.8% |
| Nov 2017 to Jan 2018 | Bullish | Nov 28 | 198,540 | 39.4% | +1.4% | -1.1% |
| Jun to Nov 2018 | Bearish | Sep 25 | -83,677 | -18.2% | +4.4% | +14.2% |
| Apr to May 2019 | Bearish | Apr 30 | 76 | 0.0% | +14.6% | +33.7% |
| Sep 2021 | Bearish | Sep 28 | 30,861 | 6.3% | +9.0% | -6.3% |
| Jul to Nov 2022 | Bearish | Sep 27 | -43,094 | -9.4% | +5.1% | +8.7% |
| Feb 2023 | Bearish | Feb 28 | 26,575 | 6.2% | +0.2% | +12.8% |
| Sep 2023 | Bearish | Sep 26 | 23,438 | 5.4% | +21.7% | +40.4% |
| Aug 2024 to Jan 2025 | Bullish | Sep 24 | 219,029 | 38.8% | +16.1% | +42.7% |
What a trader could have known at the time
Nobody had the 6.46% cutoff in 2013 or 2015: it comes from all 20 years. So we also flagged a month only if it ranked in the bottom or top 10% of all readings up to then, from May 2009, with three years of history. On that basis, gold was higher a year after 24 of 31 bearish months (77%), against 65% for all months from May 2009. The edge shrinks mostly because February to June 2013 join the list, during that year’s slide: gold was lower a year later after four of those five months.
Ranked against only the past three years, the result flips: months with one of the four biggest bets were followed by gains 81% of the time, those with one of the four smallest 63%. Positioning mostly followed the trend.
The 1990s, when speculators betting against gold were right
The CFTC’s legacy futures-only data go back to 1986, and its “non-commercial” group included “professional money managers (CTAs, CPOs, and hedge funds)” and other speculators. Its 49 most bearish months, with net shorts of 17.84% of open interest or more, all fell between 1988 and 2001, while gold’s monthly average slid from $476.58 in January 1988 to $256.08 in July 1999. Gold was higher a year later after only 21 of them (43%), against 61% for all months since 1986. The same data’s most bullish months (net longs of 44.83% or more) were followed by gains 88% of the time.
Academic work also finds speculators following prices more than calling turns. A 2012 study of gold, silver and platinum in Resources Policy found “trader’s net positions do not lead market returns in general,” with “some evidence” that extreme positions forecast returns. A 2015 study of precious metals could not confirm “any short-term impact of speculators on returns” after 2006, but could not rule out a longer-term one. In agricultural futures, Changyun Wang found that “large speculator sentiment forecasts price continuations.”
Where hedge funds stand in gold now
On November 3, 2026, managed money held a net long of 103,614 contracts in COMEX gold, 26.0% of open interest, not an extreme by this study's cutoffs. That is higher than 55% of weekly readings since 2006 but lower than 68% of those in the past three years. Hedge funds weren’t crowded before gold’s January 30, 2026 crash: the January 27 report showed 118,159 contracts, 24.2% of open interest. The broader legacy measure looks more stretched, with non-commercial traders net long 45.0% of open interest, inside that series' top 10%. The gold price averaged $4,105 an ounce in October 2026, 18% below February’s record monthly average of $5,020.
Bullion Trading research methodology
Positions are the CFTC’s Disaggregated Futures Only report for COMEX gold (code 088691), read from its public reporting site on November 2, 2026; the latest-report figures update weekly, the study stays as computed. Each month uses its last weekly report. The extreme groups are the bottom and top 24 of the 244 month-end readings. The CFTC built the 2006 to 2009 history from later classifications, which “diminishes the data’s accuracy as it goes further back in time,” though it calls it “reasonably representative.” Counts that add options or other speculators differ: Saxo Bank reported a record net long of 292k lots in September 2019.
Gold prices are the World Bank’s Pink Sheet monthly averages (a spot average of daily rates from June 2025, the London afternoon fix before), at full precision through January 2026 from the February 3, 2026 release and in whole dollars after. Changes run from the average of the month after the reading to 3, 6 or 12 months later; from the reading’s own month, 21 of 24 bearish and 16 of 24 bullish months were higher a year later.
Because extreme months cluster, we shifted their list to each of the 231 possible calendar starts: 4% of shifted lists matched the bearish 92%, 37% its median gain and 65% the bullish 63%. The most bearish 5%, 15% and 20% of months were higher a year later 100%, 89% and 84% of the time, the most bullish 50%, 68% and 67%. Counting contracts instead of shares gives 23 of 24 bearish and 18 of 24 bullish months higher. The legacy check uses the Legacy Futures Only report from January 1986, which came out less often than weekly until 2000. Full dataset, one row per month: CSV download.
What the COT report can tell a gold buyer
Since 2006, the moments when hedge funds had given up on gold were usually followed by higher prices; their most confident moments told you little. But 10 episodes is a small sample, the 1990s ran the other way, and today’s extreme can look ordinary after a few more years of data. For someone buying physical gold to hold, positioning is useful context and a poor timing tool. Several of the best years here began in crowded months, one reason many buyers spread purchases over time.