Why is gold falling is the search, and the usual answer is the wrong one. A war still underway should have been a bid for the metal people buy when they are frightened. It has been a bid for oil instead. Oil feeds inflation expectations, those expectations feed the long end of the Treasury market, and gold pays no coupon, so it has to beat that yield to stay in a portfolio. On Friday, Oct. 2, the yield won the day. Reuters reported spot gold at $4,140.06 an ounce by 2:33 p.m. Eastern, down 0.9% on the session and about 3.4% on the week. COMEX gold futures settled 1% lower at $4,162.30. The weak September jobs report had already been published. The rally it produced did not survive the close.
The 6% figure in the headline is a futures number, and it is not the same market as that $4,140 spot print. COMEX gold futures closed September at $4,186.70, down 6.6% from the Aug. 31 close of $4,481.50. Reuters, writing on Oct. 1, described September as a drop of over 6%. From the Aug. 24 futures close of $4,697.80, the highest daily close in this series, the Oct. 2 settle of $4,162.30 is 11.4% lower. I am not pasting a 7.5% monthly drop onto $4,140. On Sept. 26, when spot was $4,286.20 at 08:59 UTC, FinanceFeeds wrote that gold was down 7.5% against where it fixed a month earlier. That sentence was about the fix, at a higher price, ten days before Friday’s close. It is not this week’s tape.
Key facts
- Spot gold was $4,140.06 at 2:33 p.m. Eastern on Oct. 2, down 0.9% on the day and about 3.4% on the week. COMEX futures settled 1% lower at $4,162.30 after an earlier gain of more than 1%. Source: Reuters, via Kitco.
- COMEX futures fell 6.6% in September, from $4,481.50 on Aug. 31 to $4,186.70 on Sept. 30. The Oct. 2 settle is 11.4% under the Aug. 24 close of $4,697.80. Source: Yahoo Finance GC=F daily closes, retrieved Oct. 4. Futures, not spot.
- US employers added 29,000 jobs in September, against a Reuters poll of 90,000. August was revised to 133,000 from 162,000. The unemployment rate rose to 4.2% from 4.1%. Sources: Reuters and Reuters on the Fed.
- After the jobs data, CME FedWatch put the chance of an October rate hike at 22.7%, down from 24.4% the prior session and 64.2% a week earlier, Reuters reported in its equity-close story. The next Federal Reserve meeting is Oct. 27–28. Source: Reuters.
- HSBC cut its average gold forecast to $4,490 an ounce for 2026, from $4,560, and cut its 2027 forecast to $4,825, citing further Fed hikes and higher oil as near-term pressure. Source: Reuters, Oct. 1.
- Nicky Shiels at MKS PAMP told Reuters the post-2022 premium over what real yields and the dollar would justify is roughly $840 an ounce, up from about $120 before 2022, and “is not fading.” Source: Reuters, Oct. 1.
The Friday test was the jobs report, and gold failed it
The setup into Friday was the one gold bulls wanted. A hot payrolls number would have raised hike odds and pushed the metal lower. A soft number was supposed to do the opposite. Kyle Rodda, senior financial market analyst at Capital.com, put the hot case on the table before the release. “If it comes in hot, it may increase the chances of rate hikes from the Fed. That would potentially weigh even further on gold prices,” he told Reuters. The number was not hot. Payrolls rose 29,000 against a Reuters poll of 90,000. August was revised to 133,000 from 162,000. Unemployment rose to 4.2% from 4.1%.
The first reaction matched the textbook, and the close did not. Reuters said bullion gained more than 1% after the report, then reversed. Spot finished at $4,140.06, off 0.9%. Futures settled at $4,162.30, down 1%, not at the bounce. On this futures series the Sept. 25 close was $4,321.20, so the week into Oct. 2 was a 3.7% decline. That is a close-to-close futures week. Reuters’ 3.4% is a spot week. They should not be averaged. They point the same way.
Rate odds moved the way a weak report is supposed to move them. CME FedWatch, in Reuters’ equity-close story, put an October hike at 22.7%, from 64.2% a week earlier. The next meeting is Oct. 27–28. The gold headline was the weekly drop, not the cut in hike odds.
That is the new fact relative to the last FinanceFeeds gold piece. On Sept. 26 the spot print in that article was $4,286.20. Friday’s Reuters spot print of $4,140.06 is 3.4% under that timed reading. They are two spot observations, not one official fixing series, and the gap is still the point. On Sept. 23, with gold near $4,301, an October hike was close to a coin flip. By Friday the hike was the minority bet, and the price was lower anyway. The path from the late-August futures high did not need a hotter labor market to continue.
“Gold bugs are perhaps refusing to get carried away for the time being, knowing that the Fed retains a hawkish bias,” said Han Tan, chief market analyst at Bybit, in the Reuters close.
Why a war bid became an oil bid
Gold’s competition is not the overnight rate. It is the yield an investor gives up by holding a bar instead of a bond. Przemyslaw Radomski, in an FXStreet note dated Oct. 2, put it in one sentence. Gold pays no interest, so for a long-term holder the relevant yield is the 10-year or the 30-year, not the Federal Reserve’s policy rate. A softer Fed helps the metal only if the long end follows. This week, he wrote, it did not.
The note still treats the jobs report as unreleased. He writes that it “comes out today at 8:30 a.m. Eastern,” and that gold is near $4,214. The page stamp is 17:53 GMT, 1:53 p.m. Eastern, after the release. The prices in the note are the pre-release tape, not the $4,140 close. On his figures the dollar index closed Thursday near 102.1, its highest since April 2025. The 10-year touched 5.344% on Thursday before closing at 5.248%, and the 30-year traded above 5.66%. A spike to $4,251 on soft inflation was sold within hours. Thursday’s $15.60 rise in gold, to $4,202.30, came with a 1.21% drop in the VanEck Gold Miners ETF, which he read as the tell. From there he put the next stage of his October sequence at a first target near $3,920.
That $3,920 is his target, not a FinanceFeeds forecast, and the note had not absorbed Friday’s close. The mechanism held anyway. Hike odds fell, the early gold gain was sold, and the long end had already done the week’s work. In the same piece he put Thursday’s Brent jump at 4.4% to $102.31, and wrote that the Pentagon was sending the USS Theodore Roosevelt and 10,000 sailors and Marines toward the Gulf. His chain runs from oil to inflation to the long bond to the dollar. The war is inside the gold price as a higher yield.
Amy Gower, commodities strategist at Morgan Stanley, made the portfolio version of the same point to Reuters on Oct. 1, with the metal still above $4,000 and the yield already the problem. “For a non-yielding asset, all else equal, that is going to be challenging because gold has to compete against that yield level in people’s portfolios,” she said. She added that a price above $4,000, with yields that high, is evidence of demand that is not the yield. The Friday settle did not take futures through $4,000. It did take them to $4,162.30, and it took spot to $4,140.06. The competition she described is the reason the jobs-day bounce failed. The same yield move was already the silver story on Sept. 26, when the 10-year had touched 5.22% and gold was still near $4,280.
“The real-yield framework still anchors fair value, but since 2022 a persistent, structurally higher premium driven by reserve diversification and geopolitical hedging has become the dominant driver that macro factors no longer capture, and it is not fading,” said Nicky Shiels, metals strategist at MKS PAMP, in that Oct. 1 Reuters analysis.
The $3,900s are a chart conversation, not a bank target
The $3,900 talk is real, and it is not the same thing as a house forecast. On Sept. 29, Canarinho, who describes himself as a family-office manager, wrote on X that gold was carving a descending triangle. “A breakdown of the $3,900 support would give shorts a reason to press,” he wrote. “Longs, in the other hand, will almost certainly defend the $4,000/$3,900 zone first.” The post is here. It is a chart opinion from a week before Friday’s close. It is not a settlement, and it is not what HSBC published.
James Stanley, senior market analyst at FOREX.com, was more specific about the nearby levels and less specific about $3,900. On Sept. 28 he called a descending-triangle breakdown and listed $4,100 to $4,104 as the next support, then $4,000 as “the big spot,” with resistance stacked from $4,200 up through $4,300 to $4,320. That post is here. Futures then closed Monday, Sept. 28, at $4,168.40, and Friday at $4,162.30. His $4,100 area was not the Friday settle. Spot, at $4,140.06, was closer to it than the futures close was. A Sunday open is not in this article. Oct. 4 is a Sunday, the cash market is shut, and there is no weekend print to add.
Those three figures are not one target. Canarinho’s $3,900, Radomski’s $3,920 and Stanley’s $4,000 are three sentences about where selling might be tested. None of them is the Oct. 2 settle. The bank language points at $4,000 as a floor people are naming, not as a level that has broken. Around Sept. 30, coverage of Gower’s CNBC appearance said she still favored gold on a 12-month view and called $4,000 a strong floor. That is a comment, not a dated Morgan Stanley target. HSBC’s $4,490 and $4,825 figures are averages. An average above Friday’s spot price is not a statement that Friday was the low.
I am not rewriting the Sept. 26 scenarios. At a spot reading of $4,286.20 that piece set a bull case at $4,750, a base at $4,300 and a bear case at $3,850, 10.2% under that morning’s price, if real yields held at or above the 2.85% close of Sept. 24. Friday’s print was a lower spot price alongside a lower hike probability.
| Level | Who said it | What it is |
|---|---|---|
| $4,162.30 | COMEX settlement, Oct. 2 | Futures close. Down 1% on the day. Not a forecast. |
| $4,140.06 | Reuters spot, 2:33 p.m. Eastern, Oct. 2 | Spot, down 0.9% on the day and about 3.4% on the week. A different market from the futures settle. |
| $4,000 | Gower, late September; Stanley, Sept. 28 | A floor she called strong, and the support he called the big spot. Not a guarantee either has traded this month. |
| $3,920 | Radomski, FXStreet, Oct. 2 | His first technical target. The note still treats payrolls as ahead of it. Not updated here for the $4,140 close. |
| $3,900 | Canarinho, Sept. 29 | Chart support. A breakdown, in his post, would be the short signal. Longs, he wrote, defend $4,000 and $3,900 first. |
| $3,850 | FinanceFeeds, Sept. 26 | Published bear scenario from a $4,286.20 spot reading. Left as that scenario. |
What the buyers actually did
Shiels’s premium is why a yield model alone does not land on $4,140. Reuters reported her “debasement and de-dollarisation” estimate at about $120 an ounce before 2022, an average above $1,000 since, and roughly $840 now. A premium of that size can sit in the price while the metal still falls. From the Aug. 24 futures close of $4,697.80 to the Oct. 2 settle of $4,162.30, the drop is about $535. Her claim that the premium “is not fading” is a statement about the post-2022 bid. It is not a statement that September was up.
The physical number Reuters put next to that premium is Chinese imports. China brought in 1,077 metric tons in the first eight months of 2026, a pace Reuters said annualizes to the highest in 11 years. “Some of that is going to PBOC holdings, but a lot of that isn’t, so there must be demand elsewhere in the country,” Gower told Reuters. FinanceFeeds laid out the official-buying case in August, at a higher price. HSBC’s October cut, $70 off the 2026 average, is this week’s institutional response. It is a mark-to-market. It is not a $3,900 call.
The dollar did not hand gold that bull case. A Reuters wrap carried by Kitco had the greenback heading for a fourth straight weekly gain against the euro, with the euro last up 0.39% at $1.1285. Dominic Bunning, head of G10 FX strategy at Nomura, called the jobs figures a Goldilocks set: activity still resilient, without significant inflation pressure. A bull case that is only “hike odds came down” needed a weaker dollar. Friday did not deliver one.
What would have to change before the October meeting
Nothing in the cash market prints on Sunday, Oct. 4. Futures can reopen Sunday evening in New York. This piece does not invent that open. The dates that can change the argument are the ones in Radomski’s note: ISM services on Monday, September CPI in mid-October, and the FOMC on Oct. 27 and 28. A hotter CPI puts the 22.7% hike probability back up. A cooler CPI, with the long end still rising, is Friday again. The metal gets a headline bid and then has to live with the yield.
Three paths follow from levels other people have already published. They are scenarios, not a new price call. In the first, the long end stops making highs, spot holds above the $4,100 area Stanley marked, and futures stay above the $4,162.30 settle into the meeting. Gower’s $4,000 floor is never tested. That path needs the oil-to-yield chain to pause. In the second, Friday repeats: a soft data point lifts gold for a few hours, hike odds fall, and the close is lower because the 10-year does not follow. Radomski’s $3,920 and Canarinho’s $3,900 are then the chart levels in front of the market, and $4,000 is only a zone people said would be defended. In the third, real yields stay at the highs behind the Sept. 26 bear case and Shiels’s premium starts to shrink. That is the path back toward the published $3,850 scenario. I am not giving it a new probability. Friday loosened hike odds and did not loosen the long end. Until one of those two facts changes, why is gold falling still has the same answer. The yield is the bid the war actually produced.
FAQ
Why is gold falling during a war?
Why is gold falling is a yield question, not a headlines question. Gold pays no interest. The war has lifted oil, and oil has lifted inflation expectations and long-term Treasury yields. A higher yield is competition for a bar that pays nothing. On Oct. 2, spot gold closed at $4,140.06, about 3.4% lower on the week, even after a weak US jobs report cut the odds of an October Federal Reserve hike.
Did gold fall 7.5% this month to $4,140?
No. That 7.5% figure is from a FinanceFeeds piece on Sept. 26, when spot was $4,286.20, and it described the change against the fix from a month earlier. It is not the move to Friday’s price. COMEX futures fell 6.6% in September, from $4,481.50 on Aug. 31 to $4,186.70 on Sept. 30. Friday’s spot print of $4,140.06 and the futures settle of $4,162.30 are separate markets.
What did the September jobs report do to the gold price?
Payrolls rose 29,000, against a Reuters poll of 90,000, and unemployment rose to 4.2% from 4.1%. Gold rallied more than 1% after the release and then reversed. Spot was $4,140.06 by 2:33 p.m. Eastern, down 0.9% on the day. Futures settled at $4,162.30, down 1%. CME FedWatch, via Reuters, put an October hike at 22.7%, down from 64.2% a week earlier. The metal still finished the week lower.
Who is calling a gold break toward $3,900?
Traders, not the banks in this week’s notes. On Sept. 29 Canarinho wrote that a break of $3,900 support would give shorts a reason to press, and that longs would defend $4,000 and $3,900 first. Przemyslaw Radomski’s Oct. 2 FXStreet note put a first target near $3,920, in a piece that still treats the jobs report as not yet released. James Stanley, on Sept. 28, pointed to $4,100 and then $4,000. HSBC’s published figures are averages of $4,490 for 2026 and $4,825 for 2027.
Is $4,000 a floor for gold?
It is a level several commentators have named, not a level this article is setting. Amy Gower at Morgan Stanley called $4,000 a strong floor in late September. Friday’s spot close was $4,140.06 and the futures settle was $4,162.30, so $4,000 was not tested in that session. A floor that has not traded is a comment about where buyers might appear. It is not evidence that they already have.