Gold's best week was bought by Europe, not the Fed

Gold closed at $4,066.63 on Friday, and the week that ended there had an 80% chance of a December rate hike priced into it. Not a cut. A hike. The metal still gained 4.93% from $3,875.44 — its largest five-session advance since the week to 24 August on this desk's own Investing daily series — and it did it while a 165,000 payroll print pushed the dollar and short-term Treasury yields higher (Kitco, Reuters).

That combination should not work. Gold pays nothing; a hike raises what nothing costs. So either the market is wrong about December, or the marginal buyer of gold is no longer the person who cares about December. The World Gold Council's October ETF report, published Thursday, says it is the second one.

Two records that point in opposite directions

Global physically-backed gold ETFs took in US$2.7bn in October, a fourth consecutive month of expansion, lifting collective holdings 19t to 4,275t — another month-end record. Assets under management fell 6% to US$542bn over the same month (WGC).

More metal, less money. That is what happens when the price falls 6.3% on the LBMA PM fix and nobody leaves. AUM is passive; tonnes are active. In October the tonnes went up. September is the comparison — holdings rose 67t to 4,256t while AUM fell 7% to US$574bn (WGC). Two record tonnages, two falling AUM prints, and US$41bn plus 246t added over the first ten months of the year. That is not distribution; it is accumulation into weakness.

Europe has bought for eighteen straight weeks

The global total hides the only thing that matters — where the money was. October splits cleanly:

• Europe: +US$1.8bn, a fourth straight month, and the region has now taken net inflows for 18 consecutive weeks. The UK led, with German, Swiss and French funds all positive.
• Asia: +US$1.3bn, also a fourth straight month, led by China, with Japanese funds adding despite the Nikkei's pullback.
• North America: −US$512m, the only region with outflows — and US-domiciled funds were the entire loss, because Canadian funds kept attracting money.

The council's explanation of the European bid is the sentence worth keeping: the streak has survived gold's weakness, surging yields across the region and changing expectations for local central banks, which "suggests that demand for gold ETFs may be increasingly strategic, reflecting persistent political and fiscal uncertainty across the region."

Strategic money does not check the December FedWatch. That is the buyer that showed up in the week gold rose while hike odds did.

The North American number is the mirror image, and its cause is exact. US technology equities sold off in October on AI capital-spend worries and, in the council's words, "some investors turned to their gold holdings as a ready source of liquidity." Selling the one thing that still has a bid is what gold is for. The tell is what came next: US funds were back to mild inflows in the final week of October, when local equities plunged again. Holders who return within a fortnight have not changed their mind — that was a cash-flow event dressed as a view.

The futures crowd has cut 194 tonnes in two months

There is a second set of holders, and it has been doing the opposite. Total COMEX net longs fell 94t in October to 560t, the lowest since the end of July. Managed money cut 53t, other reportables 54t, and spreading positions came off 13t.

Put September beside it and it stops looking like noise: total net longs down 100t to 654t, managed money down 84t to 387t (WGC). From the end of August the futures market has shed roughly 194 tonnes of net length — about a quarter of the position — in two months.

That de-grossing is the mechanism behind October's fall, and the council says so plainly: Fed rate-path repricing, speculative de-grossing, and a rotation out of both equities and gold into bonds (WGC). Yields led, and gold followed them back up on Friday when they stopped rising.

In "Where floor art thou?", published the same day as the flow data, the council ran its three-condition bottoming framework: gold stretched well below its 200-day average, a dollar spike, and a speculative washout. Two of the three are in place. Positioning is not. "Speculative froth has not yet been fully cleared," the note says, and a further shake-out in futures positioning would need to happen before a durable recovery can hold.

The table behind that framework is small and the council flags it: ten months with at least two conditions in place, 90% positive three months later, median 4.2%; all three firing in five episodes, higher every time, median 9.8%. That is a three-month statement resting on ten observations, not a forecast for Tuesday.

Friday was the same fight inside one session

The tape ran this argument all week: Monday's range was $95, Tuesday $80, Wednesday $151, Thursday $89, Friday $78 — a mean of $98.57 against an ATR(14) of $83.57. Gold is trading about a fifth wider than its own recent norm.

Friday compressed it into nine hours. It opened at $4,087.48, printed $4,107.50 early, then sold for seven hours into the print. The 13:00 UTC bar ran from $4,062.62 to $4,029.07 on the release; the 14:00 bar held $4,030.18; by 18:00 it had run to $4,077.58 and the close was $4,066.63 — $37.56 repaired off the low, inside a session that still finished red.

The levels that matter into Tuesday

• $4,066.56 — the 20-day simple average. Friday closed seven cents above it, the cleanest statement available about where this market thinks it belongs.
• $4,029-4,021 — the shelf. The 22 October low was $4,020.99, the 28 October low $4,024.36, Friday's low $4,029.95: three sessions, three tests, all inside $9.
• $4,147.02 — Thursday's high and the top of the range.
• $4,177.59 and $4,198.84 — the 15 and 16 October highs, sitting on the 50-day at $4,197.59 and the 100-day at $4,202.07.
• $3,870.17 — the 30 October low, where October's fall ended.

Kitco's survey of 13 analysts came in 31% higher, 38% lower and 31% neutral, while 68% of 212 retail respondents were bullish — a split tape, which is a range traded at its edges. The week's data is Tuesday's October CPI at 13:30 UTC, then PPI on Friday (Kitco, BLS).

What would make this read wrong

One falsifier, stated in advance: if November's month-end ETF holdings print below September's 4,256t while COMEX net longs keep shrinking, the holders started selling and October was a pause rather than a bid. A break in Europe's 18-week streak is the early warning, and gold can still test $3,857 — the 38.2% retracement of the entire rise from 2015, in the council's own numbers — before any of this settles.

What the flow data supports is narrower than a floor call and more useful: the fastest seller of the past two months is closer to done than started, the slowest buyer is still buying, and price is sitting on the average it has spent five sessions rebuilding around. This desk's plan into CPI is in today's signal; Friday's payrolls reaction covers the print itself, and the ETFs hub carries the rest of the flow work.

Sources

• World Gold Council — Gold ETF Flows, October 2026
• World Gold Council — Where floor art thou?
• World Gold Council — Gold ETF Flows, September 2026
• Reuters — about 80% chance of a December Fed hike
• Kitco — Wall Street split on gold
• Investing.com — XAUUSD daily bars
• BLS — CPI release schedule

Not financial advice. Levels are reference points, not recommendations; every figure is as of the 6 November close unless stated.