Monday's inverted haven bid was an artefact of a closed bond market. Tuesday's reopening breaks it - and the morning framework predicted exactly this.

Gold took back the entire Monday decline and 40% of Monday's range in one session. It opened at 4171.37 - Monday's close to within 23 cents - was hit to 4150.68 in Asian hours, then rallied through the European open to 4200.46 before settling at 4184.16 at 17:23 GMT+8. The day's range is 49.78 dollars against Monday's 62.53. Two things in that sequence matter more than the net result. First, the Asian low 4150.68 sat four dollars below Monday's 4154.78 and was defended - a double test of the same shelf in consecutive sessions. Second, the rally was not a dollar-weakness story: the index fell 0.23% but EUR/USD printed a fresh 2026 low at 1.1150 in the same session. Gold rose while the euro made its low, which means the marginal buyer was not a broad dollar unwind.

The correction that makes this cleaner: Monday's oil move was bigger than reported. The morning edition carried WTI at 93.16, +1.43%. Two independent sources - Stnn and ATFX Capital - put Monday's WTI November settle at 93.74, +2.06%, and Stnn plus 汇通网 independently put Brent December at 106.65, +1.84%. The shock was roughly half again as large as carried. That does not soften Monday's gold decline of 0.54%; it widens the anomaly that needed explaining, and it makes today's reversal a larger move off a smaller base than the morning edition implied. Readers comparing editions should use 93.74 for Monday, not 93.16.

What this edition does not have is a New York close. At 18:00 GMT+8 it is 06:00 in New York; the equity and Treasury cash sessions open roughly 3.5 and 5.5 hours from now. Every level here is an Asia-and-Europe mark. Treating a pre-open print as a settle would repeat the error the morning edition flagged in its own methodology about carried Friday closes.

October pricing drifted back up, which is the opposite of the morning's cleanest number. The morning edition recorded October cumulative hike odds at 18.3% and read that as the market pushing the hike out one meeting. Today's readings span 17.2% to 22% - thepaper has 82.8% hold, AMarkets 80.6%, Jin10 79.5%, and one Chinese technical desk explicitly attributes the move to Hammack taking October odds from 17.7% to 22%. This edition uses AMarkets' 19.4% as the mid and discloses the range rather than picking the flattering end. December at-least-one-hike odds are 84.7% against 84.2% on Monday - essentially unchanged. The correct read is not that the deferral failed but that it stopped improving: the market took Hammack seriously without re-pricing the cycle. Goldman moved its hike expectation from October to December last week; Morgan Stanley argues the market may be overestimating the terminal rate; Piper Sandler's Kurt Lewis expects three hikes. The dispersion between desks is wider than the dispersion between market snapshots, which is unusual.

Waller speaks at 15:45 GMT+8 and this edition does not characterise what he said. He is one of the officials who argued there was no need to hike in consecutive meetings, and his remarks are the first direct test of whether the doves push back on Hammack's framework or fold to it. No verified account of his speech was available before this edition's cutoff, and a guess dressed as a summary would be worse than an acknowledged gap. The decision calendar is otherwise fixed: FOMC October 27-28, CPI Wednesday 20:30 GMT+8, and both languages - "small adjustments relatively quickly" against "no need to hurry" - will have been tested against a published core inflation number before the meeting. That is the actual deadline for the reaction-function argument to be settled.

The franc's reversal is the single most important cross-asset fact in this edition. The morning edition built a large part of its argument on the franc being the worst-performing major at USD/CHF +0.60% - used to demonstrate that the dollar was a carry bid rather than a haven bid, and therefore that an escalating war would not bid gold. Today, on a materially unchanged Gulf situation, the franc is among the strongest at about -0.24%. A one-session reversal on an unchanged backdrop means the marginal driver is the oil sign, and through it the rate path. That is the cleanest available test of whether the morning's central claim - that this cycle's gold bid requires a disinflationary shock - is a statement about economics or about positioning.

Canada is the second confirmation, and it is a negative one. With WTI down 1.49% the CAD should be the best performer on the board on a terms-of-trade basis. It is the weakest, at about -0.09%, having failed to hold the 1.4290 retracement with the 1.4300 year high overhead. This is the mirror image of Monday, where crude rose 2.06% and the CAD still underperformed. Two consecutive sessions in which the oil move failed to express in its own currency is strong evidence that crude is reaching markets through inflation expectations rather than through trade balances - which matters directly for Wednesday's headline CPI, where the energy component is the largest single contribution.

All levels here are Asia-and-Europe marks. Only two pairs - USD/JPY and USD/CHF - are derived by applying a published percentage move to Monday's published close, because no direct Tuesday fixing exists at this hour; both are flagged in the table and the yen in particular carries a 44 pip source spread. Treat this table as directional, not as fixings.

Kazakhstan breaks the central-bank argument in a way no price print can. The National Bank of Kazakhstan has begun a gold reduction programme, taking gold to 75% of total reserves by year-end - a 3 percentage point cut - from current holdings of 386 tonnes. A single reduction does not offset a 23-month PBoC accumulation, and this edition does not present it as though it does. The PBoC bought 740,000 ounces in September, its largest month of this cycle and the 23rd consecutive, taking reserves to 77.47 million ounces, about 2,410 tonnes, while total FX reserves fell 38.1 billion dollars in the same month - a deliberate reallocation, since capital left the book and gold is what it bought. The number that matters is whether Turkey, India or the Gulf funds follow. UBS's annual range of 750-1,000 tonnes is the benchmark, and 386 tonnes off is roughly a quarter of one year's buying, recurring.

Goldman's cut is a timing change, not a thesis change - and the distinction is in the 2027 number. Goldman lowered its 2026 year-end target to $4,900, attributing it to a delayed Fed easing path, while maintaining $5,400 for 2027 (FX678). What is notably absent from today's research flow is any new 2026 revision: the day's desk commentary went to oil, to diesel supply, to OPEC's downgrade and to the bank earnings ahead, not to gold targets. The panel carried forward therefore still spans Macquarie at 3,050 to UniCredit well above spot, and the prior edition's unreconciled conflicts - Wells Fargo at 3,600 and Deutsche at 4,600 against 4,900-5,100 and 5,000 in the earlier edition - are still shown and still not averaged.

One genuinely new macro input arrived today: the IMF's October World Economic Outlook raised 2026 global growth to 3.1% on AI investment offsetting the energy shock, kept the US at 2.3% for both years, cut the euro area to 1.1% and 1.2%, warned that a Hormuz disruption could take 0.9 percentage points off world growth, and now forecasts Saudi Arabia contracting 3.3%, a 5 point downgrade from July.