Gold prices today sit near the bottom of a long giveback. On the morning of October 7, 2026, spot gold traded around $4,096 an ounce, the lowest since August 5, down about 1.6 percent on the day. January's record was above $5,500. A metal that made that high and then lost roughly a quarter of it is not "doing fine if you zoom out." It is underperforming the story that got people in. The story had a name. People called it the debasement trade. Robin J. Brooks is one of the economists who both believed the story and explained why the price stopped obeying it.
Here is the only idea worth keeping. Gold is underperforming because the buyers who extended the rally do not act like owners of a haven, and the rate path that was supposed to be a scare is still the path the market is pricing. Brooks put that in three factors. New holders made gold trade like a high-beta asset. A hawkish reading of the Fed under Kevin Warsh lifted real interest rates, which weigh on a metal that pays nothing. Reserve managers, watching gold fail its safe-haven test, turned more careful, and the dollar looked safer. He later said those forces would fade and the trade would resume. The gold price decline into this week says the fade has not arrived. Recovery is a claim about those three factors. It is not a date on a calendar.
Nothing here is a recommendation to buy or sell gold, a fund, or a gold mining stock. Brooks speaks for himself. A factor is not a target.
Underperforming compared with what
The phrase needs a yardstick, or it is just a mood. Gold can be down on the day and still up over two years. It can be up over two years and still a bad haven in the week you needed one. Why is gold underperforming depends on which test you mean.
Against its own high, the failure is obvious. From a January peak above $5,500 to a print near $4,096 is a gold price correction on the order of a quarter, depending on the exact high in your data set. Against late summer, when the metal was near $4,700, the drop into early October is smaller and still large. Against the end of August, New York futures fell 6.4 percent in September alone. That is underperformance versus the recent past.
Against the job it was bought for, the failure is older. Brooks argued in June that gold fell with the stock market when the Iran war stressed markets in March. A haven is supposed to do the opposite. A metal that drops when fear rises is not doing the job. It is doing the job of a risk asset. That is the sense in which Robin Brooks gold commentary cut deepest. The price can be high and the function can be broken.
Against the dollar, he argued the same point from the other side. In a June 4 post he wrote that it was gold's safe-haven status that had been dinged, not the dollar's. Since the war, he said, the dollar had been the better refuge. If your gold investment was a bet that the dollar was finished as a reserve asset, and the dollar was the thing that rose when people were scared, the bet underperformed even when the ounce was still well above where it started the decade.
Against his own later call, the score is still open but no longer flattering. On September 26 he wrote that two catalysts were building and should help gold and the other precious metals take off. The October tape did not take off. It made a two-month low. A forecast can be early. It cannot be cited as a description of what already happened.
Who is speaking, and what he actually wrote
Robin J. Brooks is a senior fellow at the Brookings Institution. Before that he was chief economist at the Institute of International Finance and chief foreign-exchange strategist at Goldman Sachs. He writes on his own Substack and posts on X. The views are his. They are not a central-bank release and not a house view of this publication. Treat them as one informed map of factors affecting gold prices, and check the map against the price.
The map has dates, and the dates disagree with each other in a useful way. On June 4, 2026, he argued gold was not displacing the dollar. On June 26 he listed three reasons the fall had been so sharp. On August 30 he said a hawkish turn from Warsh, and even a September hike, should not be bearish for gold, because debt was out of control and real long-term yields might be capped. On September 26 he said the metal had hung on better than the hawkish shift deserved, and that the next move should be up once the market stopped pricing a long hiking cycle and once oil fell. Read them in that order. The factors come first. The optimism comes later. The price, this week, still answers to the factors.
The bubble he says came before the fall
Brooks did not start from the idea that gold was a fraud. He wrote that he was sympathetic to the debasement trade. He shares the view that fiscal policy in much of the G10 is out of control. The worry, in his telling, was warranted. The size of the move was not.
In the June 26 post, "Why Has Gold Fallen So Much?," he measured the rally from Chair Powell's Jackson Hole speech on August 22 of the prior year. By late January 2026, gold was up more than 60 percent from that speech. Silver was up about 200 percent over the same span. He called that nuts. Something that began as concern about fiscal policy had become a speculative bubble. By the time he wrote, gold was up only 21 percent from that Jackson Hole speech, and silver only 52 percent. The bubble, in his words, had corrected. It had not gone back to zero. A 21 percent gain from a famous speech is not a bear market against the old world. It is a large giveback against the mania.
That distinction is the start of an honest gold price forecast. A forecast that says "the thesis is alive" can be true while a forecast that says "the January price is the real price" is false. Brooks held both ideas at once. The thesis, for him, was fiscal excess. The January price was speculation on top of the thesis. Why are gold prices falling, in that frame, is not "because the debt got smaller." The debt did not get smaller. The extra buyers left.
Factor one: the new buyers made gold high-beta
The first of his three reasons is about who showed up. The scale of the rally, he wrote, sucked in a lot of retail investors. They are more skittish than the prior holder base. Gold began to trade like a high-beta asset. It fell when the Iran war escalated. It did not sit still and look like insurance.
His chart for this point put the gold price against the S&P 500. In March, when it mattered, the two fell together. That is a new behavior if your memory of gold is 2008 or March 2020, when the metal's path and the stock path were not the same trade. A high-beta gold price is a gold price that amplifies risk instead of absorbing it. The people who bought the top were not central banks with a decade-long mandate. They were holders who could sell in an afternoon. When they sold, the safe-haven story sold with them.
This factor is still the cleanest explanation of why gold underperformed in the moments it was supposed to shine. A war, a shock, a risk-off week: the old script says buy the bar. The new script, in his March evidence, said sell what you bought last month because the loss hurts. High-beta is not a permanent law of chemistry. It is a description of a holder base. If those holders are washed out, gold can stop trading like a speculative token. Brooks himself said it still had a long way to go before it was "back to its old self," and that it would need more washouts before it stopped trading like bitcoin. A washout is not a forecast of the next low. It is a statement that the weak holder is still in the price until the price proves otherwise.
October does not retire this factor. A slide to a two-month low, on a day when the dollar and yields rose and no mine was closed, is what a skittish holder base does. It is also what a rates market does. The factors stack. They do not take turns as neatly as a list implies.
Factor two: real rates, and a Fed the market decided had changed
The second reason is the old one, and he does not pretend it is new. Gold moves inversely with real U.S. interest rates. Real rates are what cash pays after inflation. When they rise, the opportunity cost of a bar goes up. When they fall, the bar gets easier to hold. On June 25 he wrote that markets thought the Fed's reaction function was shifting in a hawkish direction under Warsh, and that this, plus a drop in oil that pulled inflation down, was driving real rates up. Higher real rates and a whiff of disinflation are the opposite of a debasement hedge. Fiat money looks better. Gold looks worse.
He was specific about the week he was explaining. The fall over the prior week, he said, was the real-rate move after the market read a Federal Reserve meeting as hawkish. There is nothing surprising, in his words, about gold falling when real rates rise. The surprise would be if it did not.
Then he changed the emphasis, and the change is part of the record. On August 30, after Warsh gave a hawkish Jackson Hole speech and markets marked up the odds of a September hike, Brooks wrote that in normal times this would hurt gold. These were not normal times, he said. A rise in the real 10-year yield tends to push gold down. But if a new accord between the Treasury and the Fed was emerging to cap longer-term yields, the hawkish talk would not be bearish. It would cap the upside for real rates. Out-of-control debt, he wrote, was driving everything, and it would drive gold higher. A September hike, in that post, was not a bearish signal.
The hike happened. On September 16 the Fed raised the policy range, to 3.75 to 4.00 percent. By September 26 Brooks was looking at the market's addition of about 75 basis points of hikes through the end of 2027 since the July 29 meeting. That shift, he wrote, should have hurt gold. Gold had instead hung on to some of its gains since July 29. Brent crude, up about 30 percent over the same stretch, was another weight that had not knocked the metal down as a simple model would predict. He took that as evidence of conviction in the debasement trade, not as evidence that the model was dead.
His two catalysts, in that September 26 post, were a bet against the market's hike path and a bet on oil. The inflation picture, he said, did not call for a full hiking cycle. There was no way the Fed would hike as much as was priced. And oil was due to fall, in his view below $100, because tanker traffic through the Strait of Hormuz had recovered and the East-West pipeline was back, which he read as Iran losing control of the strait. Markets at that point priced about 70 percent odds of a 25-basis-point hike. He was highly doubtful. If the Fed did hike, he wrote, that would be the end of the cycle. One way or another the Fed would disappoint the hawks, oil would come down because no war lasts forever, and the debasement trade would take off again.
That is a clear gold price prediction. It is not a number. It is a sequence. Hike odds too high. Oil too high. Both fall. Gold rises. Test it against October 7, not against the elegance of the sequence. The September hike was already in the past. By midmorning on October 7, Reuters put the odds of a further hike in December near 84 percent, up from lower readings earlier that same morning. The dollar index was up 0.7 percent. The 10-year yield was at a more-than-twenty-year high. Spot gold was at a two-month low. The scare he called a scare was still being priced as a policy. Real rates, not the debt stock, were setting the week.
None of this proves Brooks wrong about the destination. It proves he was early about the clock, and that factor two did not retire in September. A gold price outlook that quotes his catalysts and skips the tape is a brochure. A gold price outlook 2026 that says "real rates are still the headwind, and the hike path is not yet a disappointment to the hawks" is just a description.
Factor three: reserve managers, and a dollar that did the haven job
The third reason is the one that challenges the most popular gold market 2026 story. The popular story says central banks are done with the dollar and are buying gold with both hands. Brooks looked at the same chatter and called a chart misleading.
In the June 4 post he plotted the share of gold in the reserves of 62 emerging-market central banks against the log of the gold price. The share rose. A lot of the rise, he argued, was the price, not a buying frenzy. Mark a stock of ounces to a higher price and the share goes up even if you bought nothing. He wrote that the pace of buying was lower than before COVID, for example. The eye sees a stampede. The flow does not match the eye.
He then made the guess that sits at the center of factor three. Any reserve manager who watched gold trade like a risk asset in recent months would wonder about its safe-haven status. His best guess was that those managers were turning more cautious on gold, and more positive on the dollar, which had behaved more like a haven than the metal had. This is a guess. He labeled it as one. It is not a survey of 62 central banks taken the next morning. It is an inference from price action. Inferences can be wrong. They can also be the right warning when the popular inference is "official buyers will catch every dip."
Official buying did not stop. Other reporting this year has shown the People's Bank of China adding tonnes for many months, and banks such as Goldman Sachs counting a larger official bid than the reported one. Brooks is not denying that some central banks buy. He is denying that the buying is a frenzy large enough to make the price immune to real rates, and he is denying that gold has replaced the dollar as the thing reserves reach for in a shock. Both denials can sit next to a true sentence about Chinese purchases. A buyer of 20 tonnes in a month does not have to save a market that is repricing a real rate. The tonnes matter over years. The rate matters this week. Gold underperforms the "central banks have got this" slogan whenever the week is a rates week.
The dollar leg of his argument is the part investors skip because it is uncomfortable for a gold thesis. If the dollar is the better haven, then a gold investment that was sold as "the anti-dollar" is the wrong hedge for the shock you actually got. You wanted protection from a geopolitical scare. You got a metal that fell and a currency that rose. That is underperformance of a function, which is worse than underperformance of a chart, because you cannot see it if you only look at a five-year gold price.
Where his own timetable slipped
A fair reading does not stop at the June list. Brooks spent late summer arguing that the list was temporary. The August 30 post said a Warsh hike would not be the bearish event normal models expect, because debt and a yield cap would dominate. The September 26 post said gold had already refused to fall as much as 75 basis points of extra hikes and a 30 percent rise in Brent said it should. Conviction, in that telling, was the news.
Conviction is a hard thing to see in a price. Hanging on to some gains since July 29 is not the same as making a new high. It is not the same as holding the January peak. It is a relative statement. Gold could be "resilient versus a hawkish rates path" in September and still be in a gold price correction from its high. Both were true. The error is promoting resilience into immunity.
October tested the immunity. The Fed had hiked. Officials were not sounding finished. Mary Daly said more hikes depended on whether inflation pressures faded. Jeff Schmid said rates still needed to rise. Markets pushed December odds higher during the morning of October 7. Oil, which Brooks wanted lower as a catalyst, was not delivering a clean deflationary gift that week. Prices rose on Middle East risk and on a storm aimed at U.S. producing regions. His catalyst was "oil down, so real rates do not have to stay high, so gold up." The week offered "oil not cooperating, real rates high, gold down." That is factor two, back in charge, with factor one available to explain why the selling was not gentle.
He also said the debasement trade would kick back into gear because fiscal policy was still out of control, but only after the market was through the hawkish scare and after more washouts. That sentence is the responsible version of his optimism. It has conditions. The scare is a condition. The washout is a condition. Neither is a closing bell. Investors who heard "Brooks is bullish gold" and skipped the conditions heard a different economist.
Why are gold prices falling, in one paragraph that survives both clocks
Why are gold prices falling? Because a rally that outran the fiscal worry pulled in holders who sell when the screen turns red. Because the market is pricing a Fed under Warsh that is still willing to hike, and real rates have risen with that pricing. Because gold failed a haven test against both stocks and the dollar, so the marginal reserve manager has less reason to chase it this month than the slogans say. Brooks wrote the first two as mechanisms and the third as a guess. The mechanisms do not need his guess to be true. A high-beta holder and a higher real rate are enough to knock an ounce down from a bubble price. The guess matters only if you were counting on official demand to cancel the knock.
The factors affecting gold prices that are not on his list still exist. Mine supply did not surge on October 7. China's reported reserve bid did not have to vanish for the price to fall. Producer countries fencing exports, a topic of the same week, are a slow supply story. They do not set an afternoon. Brooks is useful because he refuses to let the slow story veto the fast one. The fast one is the holder and the real rate. The slow one is the debt and the official bid. Gold market outlook work that uses only the slow story will be surprised every time the fast story shows up. That surprise is what this drawdown has been.
Will gold prices recover
Will gold prices recover? Brooks' answer is yes, with a wait. The debasement trade returns, in his June close, once the hawkish scare is over and once gold has washed out the holders who make it trade like a speculative token. His September answer was more urgent. The Fed would not deliver the hikes in the curve. Oil would fall. Those were the catalysts. If he is right about the path of policy, the real-rate headwind eases and factor two flips. If he is wrong, and December brings another hike and the curve keeps adding them, factor two stays a weight. The gold forecast 2026 is mostly a forecast of that fork. A gold price forecast 2026 that skips the fork and prints one number is a slogan with a decimal.
Other maps put numbers on the fork without settling it. Goldman Sachs, in notes cited earlier this week, cut a year-end 2026 fair value to $4,650 from $4,900 and kept $5,400 for December 2027, with official buying doing most of the work in the model. Société Générale's technical desk had gold pressing an interim level near $4,095, with $4,000 and the summer troughs near $3,940 to $3,960 below if it cannot retake about $4,225. Those are scenarios. Brooks did not publish a target that matches them. Do not staple his factors to someone else's price and call it his call.
A recovery that counts, on his own logic, has to show up in behavior, not in a headline. Gold has to stop falling with stocks when the next shock hits. Real rates have to stop rising, or gold has to rise despite them, which would mean a different buyer has taken the other side. The dollar has to stop being the cleaner haven, or gold has to rise for a reason other than "anti-dollar." Until one of those three changes, a bounce is a bounce inside the same regime. Tuesday's bounce, given back on Wednesday, was that kind of bounce.
There is a version of recovery he would not sign, and it is the one retail letters often sell. It says the debt is large, so the price must go back to the January high on a short fuse. He called the 60 percent run nuts. Wanting it back on a timetable is how the bubble restocks itself. The gold price prediction that matches his writing is duller. The fiscal worry is real. The price got ahead of it. The path back is a scare that ends and a holder base that changes. Neither has a closing price printed on it.
What gold mining stocks do with this
Gold mining stocks are a geared claim on the gold price, plus a claim on costs, grades, and the honesty of a mine plan. When gold trades high-beta, the shares trade higher-beta. September showed it. Futures fell 6.4 percent. The large miners in a global ranking fell 12.7 percent. Not one of the fifteen names rose. The VanEck Gold Miners fund, GDX, went from about $101 in early September to the high $80s by October 6. If Brooks is right that the metal itself has picked up the habits of a risk asset, the miners are not a shelter from that habit. They are the habit with a multiplier.
The multiplier does not know about his third factor. A reserve manager turning cautious does not call a company and cut a dividend. The share price just falls more than the ounce. Investors who bought miners as "gold plus a business" in order to avoid the metal's volatility discovered that the business levered the volatility. That is not a reason the shares are cheap. Cheap is a word for the gap between price and value after you have checked costs and guidance. Kinross cut ounce guidance for 2026 and 2027 in late September for operating reasons, not for a Fed speech. A rates-driven gold price decline on top of a guidance cut is two problems. Brooks explains the first. He does not explain the second. Do not ask him to.
If his catalysts arrive, the same multiplier works upward. A metal that retakes lost ground can lift intact producers by more than the ounce. That is not a reason to own them while factor two is still the week's driver. It is a reason the rebound, if it is real, will look like insight and will have been leverage. Leverage is not a thesis. The thesis is still the three factors. The shares are a noisy way to express them.
How to use the factors without turning them into a slogan
A factor is useful if it can be falsified. Brooks' list can.
High-beta fails as a description if, in the next risk-off week, gold rises while stocks fall. One week is not a new regime. A string of them is. Until then, why is gold underperforming its haven job has an answer. The holders are wrong for the job.
The real-rate factor fails if gold rises while real yields rise, for more than a headline day. That would mean a buyer who does not care about carry has taken control of the margin. Official demand is the candidate. It has to show up as tonnes and as price, not as tonnes alone. Tonnes with a falling price mean the official bid is real and not large enough for the week.
The dollar factor fails if the next scare lifts gold and sinks the dollar together, the way the debasement trade promised. If the dollar rises and gold falls again, he called the haven ranking correctly, at least for that scare. Gold investing built only on "the dollar is done" is then still the wrong book.
His timetable fails, and has already slipped, if hike odds stay high and oil does not deliver the drop he wanted, and gold keeps making lower lows. That failure does not erase the factors. It erases the claim that they were about to flip in late September. The gold outlook 2026 that survives is the one that keeps the factors and throws away the calendar.
The close
Robin J. Brooks did not say gold was finished. He said a warranted worry about fiscal policy became a bubble, and the bubble corrected for three reasons. The new buyers were jumpy, so gold traded like a high-beta asset and failed as a haven when war stress hit. The market decided the Fed under Warsh had turned hawkish, so real rates rose, and gold did what it usually does when real rates rise. Reserve managers, he guessed, noticed, and the dollar did the haven job gold did not do. He later said the hawkish scare would not last, oil would fall, and the trade would resume because the debt had not been fixed.
Why are gold prices falling into October? Because the scare is still in the price. December hike odds were near 84 percent on the morning gold made a two-month low. The dollar was up. Yields were at a multi-decade high. Will gold prices recover? On his conditions, only if that path breaks and the holder base stops treating the bar like a speculative token. Until then, gold is underperforming the story, not the arithmetic. The story wanted a haven and a short path back to the mania. The arithmetic wanted a real rate. This week the arithmetic won. The debt is still there. It does not get to skip the rate.
A note on sources and limits
Brooks' three factors are from his June 26, 2026 Substack post, "Why Has Gold Fallen So Much?" The 60 percent gold gain and roughly 200 percent silver gain are his measures from Powell's August 22 Jackson Hole speech of the prior year to the late-January 2026 peak. The 21 percent and 52 percent figures are his measures of what was left of those gains when he wrote. The high-beta point uses his comparison of gold and the S&P 500 in the March stress. The real-rate point matches his June 25 post on X and the June 26 essay. The reserve-manager point is his guess, labeled as one.
The dollar argument and the claim that emerging-market gold reserve shares overstate buying because of price, and that the pace of buying was lower than before COVID, are from his June 4, 2026 post, "Gold is NOT displacing the Dollar." The August 30 post, "Will a September Hike Hurt Gold?," is the source for his claim that a Warsh hike need not be bearish if longer-term yields are capped. The September 26 post, "The Debasement Trade under Kevin Warsh," is the source for the 75-basis-point shift in hike pricing through end-2027 since July 29, the claim that Brent was up about 30 percent over that span, the two catalysts, the roughly 70 percent odds he cited, and his doubt that the Fed would deliver the full path. Those posts are his opinions. They are not forecasts endorsed here.
Market facts around October 7, 2026, are separate. Reuters reported spot gold near $4,096.13 at 9:20 a.m. Eastern, down 1.6 percent and the lowest since August 5, with the dollar index up 0.7 percent and the 10-year yield at an over-two-decade high. December hike odds near 84 percent are the CME FedWatch reading Reuters cited at that hour. Earlier the same morning the odds were lower. The September 16 hike and the 3.75 to 4.00 percent range are the policy decision. Daly and Schmid's remarks are from same-day reporting. Goldman's $4,650 and $5,400 figures, and Société Générale's $4,095, $4,225, $4,000, and $3,940 to $3,960 levels, are from those firms' notes as cited in early-October reporting. They are not Brooks' levels. September's 6.4 percent drop in New York gold futures and 12.7 percent drop in the large gold miners are from mining.com's October ranking. GDX's move from about $101 on September 3 to the high $80s by October 6 is from published quotes.
This is not investment advice, not an offer, and not a solicitation to buy or sell any security or commodity. Gold and gold mining stocks can keep falling. A factor can be right and the timing wrong, or the timing right and the factor incomplete. Readers should read Brooks' posts themselves, check the live price and the Fed's own record, and speak with a licensed adviser before any decision.