October CPI preview: one tenth, and no annual rate

Gold starts Tuesday at $4,140.86, $8.35 under the Asian session high, sixteen hours from a US inflation report that arrives without the number everyone quotes.

The figure that will not print

The Bureau of Labor Statistics will omit the 12-month percent change for both headline and core CPI. It never collected October 2025 prices during last year's lapse in appropriations and could not reconstruct the gap; a BLS spokesperson, Cody Parkinson, confirmed the agency cannot calculate the annual change for either measure, and the real-earnings release also goes unpublished (Quartz, citing Bloomberg, Bloomberg). The last official annual rates are September's 3.7% headline and 2.6% core, up from 3.4% in August.

That absence is the story. The annual rate is the part of CPI that becomes a headline, and this month it does not exist. What is left is a one-decimal monthly change and a Fed decision on 9 December.

Into the hole the aggregators have poured model output. FactSet publishes a median 4.0% headline and 2.8% core year-over-year — from four estimates spanning 3.90% to 4.00% (FactSet). The Cleveland Fed nowcasts 3.74% and 2.48%, and its own page notes it substitutes a nowcast for the missing October 2025 CPI (Cleveland Fed). Both estimate a number that will not exist. If a terminal shows an annual CPI at 13:30 UTC, it is somebody's model, not the bureau's arithmetic.

What forecasters actually expect

• Headline m/m +0.3%. The Reuters poll figure and the Bloomberg survey median (CNBC/Reuters, Quartz). FactSet's survey is lower at +0.24%, Bank of America higher at +0.4% (TradingKey). Range 0.24% to 0.4%.
• Core m/m +0.2%. Bloomberg and FactSet both have it, the Cleveland Fed nowcast is +0.21%, RBC and Bank of America are at +0.3%. BLS publishes one decimal, so anything at or above 0.25% unrounded prints 0.3% and makes a third consecutive 0.3% core month.
• Gasoline does the cooling, not services. UBS estimates energy added 0.39 percentage points to September's headline and about 0.04 in October, with Brent at $94.99 on Monday, down 3.4% (CNBC). Goods pass-through, shelter and services are the core, and none improved.
• The Fed is priced. Futures imply around 82% for a 25bp hike on 9 December on CME FedWatch, CNBC's gold wrap quoted 84.5% the same afternoon (CNBC), and Kalshi's December contract traded at 77 cents through the weekend (FinancialMarkets).

The grid before the print

Monday traded $4,055.14 to $4,149.21 and closed at $4,140.01 — a $94 session inside an ATR(14) of $87.15 on the fourteen completed daily bars, Sunday stub excluded. The dollar index eased 0.37% to 101.84, the 2-year yield rose to 4.564% and the 10-year to 4.876% (CNBC).

• $4,149-4,155 — first resistance. Monday's high at $4,149.21 and 5 November's $4,147.02: two failures inside $8.
• $4,166-4,178 — second. The 14-15 October rejection shelf at $4,170.48 and $4,177.59.
• $4,186-4,199 — third. The falling 50-day average at $4,185.94 and 16 October's $4,198.84; above that, $4,217.47 from 12 October.
• $4,100-4,105 — first support. Four hourly lows on Monday at $4,105.27, $4,100.72, $4,100.66 and $4,101.44.
• $4,083-4,085 — second. Monday's London low of $4,084.78 and the 23.6% retracement of the 3,870.17-to-4,149.21 leg at $4,083.36.
• $4,055 — third. Monday's low, Friday's electronic low at $4,054.40 and Sunday's close at $4,055.73: three touches of one number.
• $4,030 — fourth. Friday's low, then $4,009.69 at the 50% retracement.

The pre-print box is $4,100-4,155; anything outside it before 13:30 UTC is positioning, and positioning is what gets reversed.

Three prints, three ladders

Hot — core 0.3%, headline 0.4%. With 82% already priced, a hot core adds real yield rather than probability. The dollar is the crowded long, so watch whether the 10-year clears 4.90-5.00%. Gold losing $4,100 tests $4,084, then $4,055; a close under $4,055 opens $4,030 and $4,010.

In line — core 0.2%, headline 0.3%. Nothing reprices. December still rests on the 165,000 payroll print, and one monthly CPI without an annual rate does not overturn it. Gold chops $4,100-4,155: fade the edges of the box, do not chase the first candle.

Soft — core 0.1%, or headline 0.2%. The only outcome with room: 60 to 70 points of hike probability to give back, against a dollar whose speculative length is already extreme. Gold reclaims $4,149-4,155, then $4,177, then the 50-day at $4,186 and $4,199.

What voids each scenario

• Hot is void if $4,100 holds on a hot core. That means the print was in the price and the dollar's length did the work.
• Soft is void below $4,084, and needs a close above $4,155 to mean anything after two rejections at $4,149.
• In-line is void if either edge breaks and holds for an hour. A box that breaks is a trend.
• The second catalyst is at 18:00 UTC: $42bn of 10-year notes auctions after CPI, and a weak sale compounds a hot print (FinancialMarkets).

Around 13:30 UTC

Gold spreads widen in the thirty seconds around the print, and the first quote is often not the first tradable price. Limit or stop-limit orders only, half the normal clip until the book refills; a stop inside the release candle is a donation, because the wick on this metal runs tens of dollars. The desk's own signal plan flattens at 13:15 UTC and carries nothing through — have a map, not an order.

Sources

• Quartz, citing Bloomberg, on the omitted 12-month changes, Bloomberg, FactSet estimates, Cleveland Fed nowcasting
• Consensus and dollar — Reuters via CNBC, gold wrap, Treasury yields, forecast dispersion, Kalshi and the 10-year auction
• Levels and ATR: the desk's own Investing.com XAU/USD daily and intraday series, pulled 9 November 2026.

Levels hold for this session only, and a monthly inflation print can move gold tens of dollars in a minute. Not financial advice. The last one is graded at the September CPI result; the series runs on the CPI topic hub, the daily grid at /levels, all coverage on the news board.