Gold Outlook: Core CPI 0.3% Three Months, Can $4,000 Hold?

Most of the messages I got this morning asked the same question: does 4,000 break? My answer: not cleanly on the first attempt. That is not a gut call. Put three things side by side and it falls out. One: the print, and the extra tenth of a percent At 22:30 on 10 November, US October CPI: +0.4% m/m against 0.3% expected, 0.6% in September. Core +0.3%, against 0.2% expected, and 0.3% in both August and September. No year-on-year figure - the BLS could not collect a comparable October 2025 base because of the shutdown. Most coverage stopped at "0.4 beats 0.3". The uncomfortable part is core printing 0.3% three months running. Annualised, that is roughly 3.6%, close to twice the Fed's 2% target. A hot headline can be blamed on oil. Three straight core prints at 0.3% cannot. Inside the report: energy +0.7% (gasoline +1%, natural gas -0.8%), food +0.5% (cereals +1%, beef +0.9%), shelter +0.3% - which the BLS says contributed more than 30 dollars to the monthly increase - transport services +0.8%, medical care commodities +0.5%, airfares +0.4%. Apparel, the tariff-sensitive line, rose only 0.1%, and used cars fell 1.2%. Oxford Economics reads it as food doing the heavy lifting: energy eased as expected, but food inflation accelerated after a flat stretch, pointing to the lagged pass-through from the energy shock into non-energy goods. Vanguard raised its end-2026 core inflation forecast to 3.2%. Two: the transmission, repricing in one night After the release: 10-year Treasury yields touched 4.94%, the 2-year moved from around 4.55% to 4.731%, a jump of 18.4bp, and DXY closed at 102.24, up 0.58%. Equities: Dow 52,231.89 down 0.30%, S&P 500 7,520.93 down 0.38%, Nasdaq 25,065.87 down 0.74%. Nvidia -5.4%, down 9.6% over two sessions. CME FedWatch: a 25bp move at the 8-9 December meeting is fully priced; 50bp odds went from zero before the data to somewhere in the 10-13.6% band, depending on the snapshot. Waller said it before the number landed: he leans toward 25bp in December, but if there is no improvement on recent months, 50bp belongs on the table, because 25bp alone puts the Fed's credibility at risk. CICC, in a note at 08:41 Beijing time on 11 November, kept 25bp as its base case, arguing that policy is not clearly behind the curve and financial conditions have already tightened at the margin. Read together, they say the same thing: 25 is the base, 50 is the tail. PPI on 13 November is the next print that can move that pricing. Three: oil is pulling the other way This is my main reason for doubting a clean break of 4,000. WTI settled at 86.44 yesterday, having traded down to 83.88, a two-month low; Brent 92.02. US officials played down the need for further military action on Iran, and crude is down roughly 15% this month. Energy carries the biggest weight and the fastest pass-through in CPI. October still showed energy +0.7%, but this month's oil move has already happened. If November energy turns negative and core stops rising, the odds of 50bp in December fall quickly. That is the one chain in this repricing that works in gold's favour. Four: the chart, double top confirmed 9 November high 4,148.60. 10 November high 4,146.696. Nineteen tenths of a dollar apart, one session between them. That is a double top. Last night's candle: open 4,140.93, high 4,146.696, low 4,013.20, close 4,032.62, down 2.62%, range 133.5 dollars. It closed below the 20-day Bollinger mid at 4,059 and has not reclaimed it. The 4,103 support I mapped in yesterday's piece was taken out and closed below. That is what a failed support looks like: not a wick and a bounce, but a close underneath. Where we are now: 4,048.74 (09:44 Beijing time, 11 November, London spot via Tonghuashun), about 16 dollars above yesterday's close. Session high 4,050.95, low 4,026.24. Below: 4,026, today's low; 4,013.20, yesterday's low; 4,000, the round number; then 3,970; then 3,858, the 38.2% retracement of the entire 2015-2026 advance, per WGC. Above: 4,059; 4,080, the heaviest resting offers on the OANDA book; 4,147, WGC's first resistance. Five: how the two sides pay Hold-4,000 case: buy near 4,000, stop under 3,970, first target 4,059, second 4,080. Entering at 4,000 with a 3,970 stop and a 4,080 exit is 80 up against 30 down, about 2.7 to 1. Break-4,000 case: do not catch the knife. Wait for a bounce into 4,080, sell there, stop 4,110, target 4,000. Same 80 against 30. If 4,000 cannot be reclaimed on a close, 3,858 is the next shelf. Both only work if you do not chase. Yesterday's range was 133 dollars; the Asian session so far is 24. That kind of low-volume retracement is exactly what gets mistaken for a reversal. Six: three things that would invalidate this First, PPI on 13 November. If it also comes in hot, 50bp pricing moves toward 30%, and 4,000 probably does not hold. Second, equities. Nvidia is down 9.6% in two sessions. A systemic tech drawdown sends safe-haven money back into gold, the opposite force to higher yields; which one wins depends on the size. Third, central-bank demand. WGC has Q3 official buying at 321 tonnes, up 42% y/y, and 675 tonnes for the first nine months. That bid does not care about levels. It is the mattress underneath, but it does not catch short-term knives. Last Are you buying above 4,000, or waiting for it to break? Give me your level and your reason - that is worth more than any forecast I can write. Tomorrow I will put the replies together into a real cost distribution. Structural read, not investment advice. No return is promised or implied.

Related

Check out more trading insights in the Trading Journal.