Gold lost 6.8% in October and the last session took half

Gold settled the month on Friday at $3,875.44 on the desk's Investing.com XAU/USD series, with Market Briefer quoting $3,884.90 — call the band $3,875–$3,885, down $146.08, or 3.63%, on the session (Market Briefer). Against September's $4,157.34 close that is a monthly loss of $281.90, or 6.78%.

Almost everything that happened across twenty-two sessions happened in three, and the last of those was the largest.

Three sessions carried the whole month

The desk's own daily series (Investing.com XAU/USD) splits October cleanly:

• 1–21 October: $4,157.35 open to $4,140.97 close — −0.39% across fifteen sessions.
• 22–23 October: $4,140.33 open to $3,995.80 close — −3.49%, or −$144.53.
• 26–29 October: $3,998.30 to $4,026.57 — +0.71%.
• 30 October: $4,023.94 to $3,875.44 — −3.63%, or −$146.08, on the widest range of the month.

Weekly closes: −0.35%, +1.24%, −0.69%, −4.07%, +0.77%. Three weeks were noise. One week was the month.

The high was $4,225.62, printed on the second session, 2 October. After that gold produced a lower high on every leg that mattered: $4,217.47 on the 12th, $4,198.84 on the 16th, $4,190.57 on the 19th, $4,155.64 on the 20th, $4,145.40 on the 22nd, and $4,082.79 on the 28th. Seven rungs down a ladder, $143 in total. The first low, $3,975.64, came on 23 October, and every rebound ran into the same descending ceiling; the month's true low, $3,870.17, came on the last session.

Oil ran October in both directions

Nothing in the inflation data did this. Core PCE printed exactly on consensus at 0.3% month-on-month and 3.1% year-on-year, and December hike pricing did not move on it: 83% before the release and 83% after (CNBC).

The 22 and 23 October sessions had a cause, and it was crude. Gold broke $4,000 on the 23rd for the first time since mid-July and lost 4% on the week — its worst week since June — because "higher energy prices fueled inflation concerns and reinforced elevated U.S. rate expectations" (CNBC). December hike odds were 78% that Friday.

That channel is mechanical, and barely about gold. Higher crude lifts the inflation path; the Fed has already tightened once this cycle into a 3.75%–4.00% range; and a metal that pays nothing loses its argument against a T-bill when the next move might be up.

The same trade reversed on Thursday and gold fell anyway

Crude has begun to come off, which is supposed to help. On 26 October gold rose as "retreating oil prices tempered inflation concerns" (Reuters). On Thursday the mechanism failed in the other direction. Brent settled 44 cents lower at $114.88 and WTI 78 cents lower at $102 on reports of US-Iran peace talks (CNBC) — a second down day, after WTI's 3% slide on Wednesday — and gold still lost 1%.

The reason is the vote. Wednesday's hold was not neutral: it carried two dissents, both in favour of a hike, from governor Lisa Cook and Cleveland's Beth Hammack, on a 10-2 split (WSJ, CNBC). The 10-year closed Thursday at 5.011% and the 2-year at 4.693%. A committee with two votes to tighten, holding at 3.75%–4.00% while the long bond yields 5%, is not a dovish Fed whatever the statement omitted.

November's question is which premium unwinds first, war or hike. October's answer was mostly the rate channel — with one session that was not.

The bid underneath never sold

Against all of that, the demand data did not crack — the strongest argument that the floor is real.

• ETFs: global gold ETF holdings rose 67t in September to a record 4,256t, with $10bn of net inflows in the month and a record $31bn across the third quarter (World Gold Council).
• Official sector: the PBoC bought 23t in September, its largest monthly addition since October 2023, lifting the third quarter to 63t and reserves to 2,410t — 8.5% of the total — on a 23-month streak (WGC China update).
• The tape: October's first two lows stopped $16 above July's $3,959.69 and $31 above June's $3,944.23 — then the final session went through both.

A market where the motivated seller is a leveraged futures position and the marginal buyer is a balance sheet that does not mark weekly behaves like this: a 3.5% two-day cascade that stopped $16 above the summer low rather than through it — until Friday. The same split showed up in the September flow data — ETFs adding metal while futures liquidated — and in the official-sector bid that held the floor on 20 October.

What November has to prove

The bear case still runs through one line: $4,060, the shelf the 22 October decline broke. Gold has not closed above it since, and Friday's close was $185 beneath it. Above that, the 20-day mean of closes sits at $4,099; a weekly close through it would be the first structural break in the ladder of lower highs since 2 October.

Below, the number was $3,975.64, and Friday broke it, through the summer base at $3,944–$3,960. From here the market is 30.7% below January's $5,595.46 high and 10.2% lower on the year — arithmetic on the desk's own series, not a forecast.

The calendar is fully dated, which makes the next six weeks easy to trade around:

• Tuesday 3 November: US midterms. Oil desks are explicitly pricing the risk that strikes on Iran resume afterwards (CNBC).
• Friday 6 November: October payrolls (BLS).
• 9 December: the Fed's final meeting of the year, with roughly an 83% chance of a hike already in the price.

The takeaways

• The oil link now cuts both ways. Falling crude helps gold only while it lowers hike odds, and the peace talks that lower crude also remove the war bid. Watch Brent's settle, not the headline.
• $4,060 settles the argument. A close above it ends the lower-high sequence. Below, the floor claim was tested at $3,975.64 and failed; the summer base is now resistance.
• Size for a $72 daily range. ATR(14) on completed bars is $72.23, about 1.9% of spot, and Friday's range ran to $156.96 — more than twice it. Stops inside $30 on this tape are noise, not risk control.
• Two dissenters for a hike is a hawkish hold. If the long case rests on "the Fed is done", the vote count says otherwise.

Update — 1 November 2026

The final session rewrote the month's arithmetic, in a way the rate channel above does not explain. Friday's −3.63% came with the 10-year yield down at 4.93% and the dollar flat, during a chip-led equity rout (Market Briefer). The 22–23 October cascade was crude and the hike path; the last session was a liquidity event that sold gold without touching its driver.

Sources

• CNBC — core PCE 0.3%/3.1%, December hike odds 83%, 10-year 5.011%
• CNBC — gold's worst week since June as oil lifted rate expectations
• CNBC — Brent settles $114.88, WTI $102 on US-Iran peace talks
• WSJ — Fed holds at 3.75%–4.00% on a 10-2 vote, Cook and Hammack dissenting
• Reuters — gold rises as lower oil eases inflation fears
• World Gold Council — ETF holdings at a record 4,256t, Q3 inflows $31bn
• World Gold Council — PBoC adds 23t in September, 23rd straight month

XAU/USD bars come from Investing.com's historical endpoints, pulled after Friday's New York close; every percentage, range and average here is computed from them, not quoted.

Not financial advice. Levels are the desk's own reading of the tape, not a recommendation, and a stop is a plan rather than a guarantee. Leveraged gold trading can lose more than the amount committed.