Gold was a passenger on the bond trade all week. At 16:00 ET the trade reversed, the passenger got off, and the driver turned out to be the macro tape after all. The 10-year settled 7.53 basis points higher at 5.1481 percent and gold settled at the low of its own session.

Spot gold settled at 4,140.52, down 24.88 dollars or 0.60 percent, and the settlement is the low of the session. The high was 4,190.27, set in the European morning, so the round trip was 49.75 dollars or 1.20 percent inside a single trading day. The prior settlement is not in dispute: 4,165.40 reconciles arithmetically with the -24.88 dollar move, and three other desks sit at 4,165.4, 4,165.50 and 4,165.18.

The shape of the day is the same one the intraday edition drew, and the ending is worse than it drew. Gold traded 4,175 to 4,185 through Asia on the Warsh interpretation, pushed to 4,190.27 before the European handover, and then broke. It was 4,146.29 at 11:00 ET with a session low of 4,142.68. From there it fell a further 5.77 dollars to the close, which means the low was still being made when the cash market shut. That matters for structure: a session that ends on its low is a session that never found a bid, as opposed to one that sold off and was defended.

What moved it was the Treasury market, and the timing is unambiguous. The 10-year traded in a narrow band around 5.07 percent through the Asian session and began a sustained climb at 20:00 Beijing, reaching 5.1481 percent at the close - 7.53 basis points on the day, with the 30-year up 6.42 to 5.4851 percent and the 2-year up 5.73 to 4.7655 percent. The equity market went the other way: the Nasdaq rose 1.11 percent, the S&P 500 0.41 percent, and the Philadelphia semiconductor index 1.61 percent to end an eight-session slide. Risk capital left duration and the metal at the same time.

Two figures from the intraday edition are now superseded and are corrected here rather than quietly replaced. The 4,146.29 mid-session quote is not a settlement and the 4,142.68 low it recorded is not the low. And the argument that the metal was being set by something other than the macro tape does not survive the last five hours of the session, in which the macro tape repriced 7.53 basis points and took the metal down with it.

The correction this file owes its readers is a list, and it is the third of its kind. Earlier October editions dropped a Beige Book release inside a stated coverage window and dropped a Hammack speech carried overnight; the 11:00 ET version of October 19 carried four table rows and omitted every Federal Reserve speaker who spoke on October 19. Williams said one more hike this year is appropriate and none is needed in 2027. Collins expects a further increase before year-end and points at data-centre construction as an inflation source. Hammack warned that persistent supply shocks can create an inflation mindset which would later require larger policy actions to dislodge. Goolsbee, Bowman and Daly argued the opposite - no urgency, wait for more data - and Waller noted that longer yields have already restrained interest-sensitive demand. Five named voices are added to this table and none of the previous four rows has been deleted.

Why the addition matters for a gold reader: the split is exactly two-sided and it maps onto the curve. The doves are talking about the next meeting and the hawks are talking about the level of rates into next year. October odds stayed near 15 percent all session, unchanged from the intraday reading, while the long end sold off 7.53 basis points. That is the cleanest possible demonstration that the market is trading the terminal level rather than the next move. The decision itself lands October 27-28, immediately before the November 3 midterms, with the quiet period running to October 29 and the personal consumption price index print on October 29.

Goldman's variance warning from the quarterly note is no longer a forecast. The firm argued that withdrawing forward guidance forces markets to lean on data and real rates, widening volatility around each release. A 7.53 basis point nominal move of which 6.95 basis points is real, on a day with no Federal Reserve release at all, is that variance arriving in one session. Waller's formulation is the one to sit with: the long end is already doing the tightening, which means the committee has less work to do at the front and the market has more reason to demand compensation at the back.

One correction carried forward from the intraday edition still stands and should not be lost in the rewrite. That edition noted it had earlier framed the 4,073.35 low as the product of an aggressively hawkish repricing, which implied a symmetric downside if that repricing returned. It did not return, the low has not been revisited, and Monday's decline still ended inside a defended range rather than breaking one. Until someone can identify what the seller is actually repricing on, calling 4,073.35 a floor remains an assertion rather than an analysis - and Monday's real-yield move is the best candidate yet, because it came from the term premium rather than from the committee.

The dollar index closed at 101.98, up 0.14 percent, having fallen roughly half a percent over the previous week. Xinhua's New York close and the three outlets that carry it verbatim give 101.980; NBD's pack gives 0.13 percent at 101.972 and Cnyes prints 101.9773. The prior session's 101.84 is the reference, which reconciles. The intraday edition held the dollar at 101.84 as a live level and argued it had not recovered 102 despite a 3.7 percent CPI and a 6.3 percent PPI. It still has not recovered 102, and after Monday it no longer needs to in order to push gold down.

That is the single most important change in this section, and it inverts the intraday argument. The 11:00 ET version concluded that a market expecting genuine tightening would have bid the dollar on hot inflation, that it had not, and therefore that gold would not respond to soft data either because both were trading the level of the policy path rather than its slope. Monday supplied the missing dollar bid and gold fell anyway, which means the metal was trading something else - and the answer, as the rest of this edition documents, is the real yield.

France produced a full sovereign repricing and the euro absorbed almost none of it. EUR/USD fell 21 pips to 1.1212 and recovered from 1.1183 intraday, while the French ten-year rose more than 10 basis points to about 4.88 percent and the OAT-Bund spread widened 8 basis points to 145-146 basis points. French debt now yields more than Italian and Greek debt, at a record premium of roughly 35 basis points over Italy, with public debt at 119 percent of GDP and about 340 billion euros of gross issuance planned for next year. Banks down hard - Wells Fargo 2.47 percent, Citi 2.04 percent, JPMorgan 1.79 percent - while the CAC 40 fell 1.11 percent. A disorderly move in one member's paper that the currency refuses to follow is country risk; it is not yet a euro-area credit event.

The franc is the position this report has to correct. USD/CHF rose from 0.8242 to 0.8265. The intraday edition held that a softening franc next to a softening metal was a de-risking signature proving the marginal buyer had abandoned gold entirely, and made that the basis of its structural caution. With the equities up 1.11 percent and the semiconductor index ending an eight-session slide, the parsimonious reading is rotation between havens and risk rather than flight from risk. That distinction matters for position sizing: a genuine de-risking regime argues for cash, while a rotation argues the metal is being sold to fund equity exposure and can be bought back the same way.

Goldman cut and UBS raised in the same week, and the gap between them is a difference of definition rather than of arithmetic. Goldman moved its year-end 2026 fair value down from 4,900 to 4,650 while keeping 5,400 for end-2027. UBS raised its year-end 2026 target to 5,600 from a prior 5,400 and argued the pricing basis has migrated from real rates to United States debt sustainability. These are not two views of one model - they are two different definitions of what gold hedges. Monday's settlement is the first clean test either side has had: the 10-year real yield settled at 2.7896 percent, up 6.95 basis points, and the metal settled 0.60 percent lower. The tape belongs to Goldman's framework. The argument still belongs to UBS's, because the debt-sustainability case is about the level of the long end rather than about tomorrow's print.

The mechanical bull case lost a day, and it is the first day it has lost one. The SPDR holding fell 0.57 tonnes to 1,058.47 tonnes on October 19, against the record 1,059.042 tonnes reported for the prior session - those two numbers reconcile exactly, which confirms the earlier reading rather than contradicting it. The iShares Silver Trust was flat at 15,478.67 tonnes. COMEX managed-money net length is 111,079 contracts, the lowest in 18 weeks, after a 3,741-contract weekly cut against 137,060 a month earlier. Silver is at 5,735 contracts, down 54 percent in a month and an eight-month low, while the Silver Institute puts the 2026 supply deficit at 46.3 million ounces. Kazakhstan's central bank has begun selling gold to rebalance reserves. A shrinking speculative book, a paused ETF bid and a widening physical deficit is the standard setup for violent two-way volatility, not for a trend.

One structural argument survives the session intact. The People's Bank of China extended its buying streak to a twenty-third consecutive month, the World Gold Council has reported a record 31 billion dollars of global gold ETF inflows in the third quarter, and positioning estimates put self-directed investor gold exposure roughly 30 percent below the 2022 peak and about 50 percent below the 2016 peak. There is no crowded long to unwind - which cuts both ways, because there is also no short base to squeeze. In a market with neither, price is set by marginal flow, and on Monday the marginal flow was a 7.53 basis point rise in the ten-year of which 6.95 basis points was real, arriving with no Federal Reserve release at all.

The range was 4,073.35 to 4,217 and gold has now closed below its pivot for the first time in three weeks. The 4,140.52 settlement is under the 4,150 five-day and ten-day band that the intraday edition named as the dividing line, and it is also the low of the session. Three published technical reads had agreed the 4-hour box would hold: the box has now been tested from the inside rather than the outside. A close below a three-week pivot matters more than its size, provided the next session does not take it back.

The decisive level is still 4,200 and it has now been rejected three times in three weeks, each time inside a window where the Fed was sounding dovish. Monday was the fourth trip to the 4,190 shelf and it failed harder than the first three, because it failed on a real-yield move rather than on positioning. Sellers do not need bad news to reject a level; they only need buyers who arrive without conviction. The buyers who lifted gold to 4,190 on a dovish headline were never the buyers who wanted 4,220, and Monday proved it by closing at the low.

Downside, the structure is genuinely thin, and Monday's real-yield move is what makes it thin. Below 4,140.52 the next shelf is 4,125, then the 4,096 to 4,073.35 band. Desk support consensus clusters at 4,141 to 4,148 and the settlement sits at the bottom of it, which means the support those desks name has already been tested today. The 10-year real yield would need to fall roughly 25 to 30 basis points from 2.7896 percent to make a 4,073.35 test comfortable. On Monday it rose 6.95 basis points instead. That is the first session in this range where the asymmetry has moved decisively to the bears, and it did so through the standard framework rather than through a crisis.

The asymmetry this week is still on the calendar, but the calendar now has a new kind of entry on it. The Fed cannot speak again until October 29. The high-impact prints are UK CPI on Wednesday, housing on Tuesday, PMIs and Michigan on Friday - and the November WTI contract expires on Tuesday, which will produce at least one two-dollar oil move that is purely mechanical. Two domestic facts already point the same way: NAHB at 29 with buyer traffic at 18, and a 30-year mortgage at 7.40 percent. Those are growth-negative facts arriving into a policy vacuum, which is the combination that produces a range breakout rather than another three-week grind - and Monday may have been the first half of it.