September CPI at 3.7%: Gold Loses the Whole Range as October Hike Odds Double

September CPI landed at 12:30 UTC on Wednesday 14 October and it was hot in the places that matter. Headline inflation ran +0.6% m/m, in line with consensus, but 3.7% y/y against 3.6% expected, and core came in at 2.6% y/y against 2.5%. Gold had spent two weeks coiling between roughly $4,143 and $4,200 in front of the number. It took two hours to lose the whole range.

What printed, and why the annual rate mattered more than the monthly

• Headline monthly matched. At +0.6% m/m the month-on-month pace was exactly what the market was positioned for. The surprise sat in the annual rate and the core.
• Energy is feeding through. Gasoline rose about 9.1% on the month, which is the sort of number that starts in one line item and ends in the services complex a quarter later.
• Core at 2.6% y/y against 2.5% expected is the piece that moved rate pricing. Core is slower-moving, it is what the Fed watches, and it went the wrong way.
• Sticky, not spiking. A single print does not reset policy. But combined with 10-year and 30-year yields sitting at 24-year highs, it removes the argument for easing into the weakness.

How the market repriced

The reaction was fast and it was in one direction:

• October hike odds roughly doubled, from about 22% to about 44%, with December still priced near 90%.
• Gold fell about 1.7% to the $4,085-$4,087 area — session lows, below the range that had held for two weeks.
• Silver fell close to 4% to about $56.99, its weakest level in roughly two months. When silver underperforms gold by that much on an inflation print, the trade is not a haven bid; it is a dollar and yield repricing.
• The dollar firmed and the 10-year held near 5.2%. Higher real yields raise the cost of carrying a non-yielding asset, which is the single cleanest explanation for gold's 2026 ceiling.

That combination — hotter inflation, higher yields, stronger dollar — is close to the worst available mix for bullion in the short term. It is also why this desk gated its playbook on the print rather than publishing a direction into it.

The honest part: what this desk did

The published 14 October playbook carried two conditional orders and a rule: no open position through 12:30 UTC. The sell zone at $4,190-$4,196 never filled — the session high was $4,166-$4,173 across five feeds, so that order expired untriggered. The buy-the-flush order at $4,118-$4,124 did fill as spot fell to $4,073-$4,078 and was stopped at $4,098 for a realised 1R loss.

That outcome is on the public ledger at /track-record, recorded the same night with the setup it belonged to. A conditional plan that produces one untriggered order and one loss is still a correct process, and the rule that kept the account flat through the release was the most valuable line in the post. The lesson worth recording is narrower: after a tier-one inflation surprise, the first approach to a support shelf is usually the wrong one to buy. The reclaim, not the touch, was the trade.

Both sides of the argument from here

The bear case, which is now the base case. Price sits below the 4H EMA9/SMA50 at $4,145-$4,155, the 21-day SMA at $4,216, the 100-day at $4,250 and the 50-day at $4,337 — a dense overhead supply stack that has capped every rally since August. Daily RSI near 31 says momentum is weak, and MACD is still below zero and below its signal. Below $4,085, the ladder runs $4,073, then $4,066 on the rising triangle trendline from last week's low, then $4,030 and $4,020. A weekly close under $4,066 resolves the triangle downward and puts $4,000 in play.

The bull case, which has not gone away. September payrolls added just 29,000 jobs with unemployment ticking up to 4.2%, and prior months were revised lower — the labour market is not obviously strong enough to absorb another hike. Central-bank buying remains firm, with purchases forecast around 720 tonnes this year on reserve diversification, and that demand does not care about a single CPI print. Stochastic is already deeply oversold, so the squeeze risk is real: reclaim and hold $4,145-$4,155 and the immediate path opens to $4,190 and $4,208-$4,214.

What decides it next: PPI, today

US PPI (September) lands with retail sales and jobless claims at 12:30 UTC on Thursday 15 October. Three ways to read it:

• Hot PPI (producer prices rising faster than expected, core sticky): confirms pipeline pressure behind the CPI surprise, keeps hike odds bid, and points gold back at $4,085, then $4,066 and $4,030.
• Soft PPI with easing yields: the first genuine counter to the CPI story. Gold squeezes into $4,145-$4,155; a 4-hour close above that zone is the reclaim signal, with $4,190 and $4,208 as targets.
• Mixed print (soft headline, firm core): the noisiest outcome, and the one most likely to produce a sweep of both sides of the immediate range before a direction establishes. This is a stay-flat scenario until the first 15 minutes settle.

Layered over all three: retail sales tells you whether the consumer is still absorbing higher prices, and jobless claims is the earliest weekly read on whether the payrolls miss was noise or trend. A soft claims number alongside hot PPI would be genuinely ambiguous — and ambiguity is a reason to trade smaller, not to trade more.

Takeaways

• The range is broken, not extended. Gold is trading below a two-week range with every major average overhead; the burden of proof is on the bulls.
• $4,145-$4,155 is the level that matters. Broken support, EMA9/SMA50 confluence, and the line that decides whether this is a trend leg or a retracement.
• The floor is a ladder, not a line: $4,085, $4,073, $4,066, then $4,030.
• The repricing is about yields, not fear. At a 5.2% 10-year, gold is fighting its own cost of carry; that is why an inflation print that would once have been bullish is currently bearish.
• Nothing goes into the 12:30 UTC release. Yesterday demonstrated the cost of being positioned into a tier-one print — the loss was small because the rule was respected, not because the call was right.

Sources

• Kitco — Gold price drops to session lows as US CPI heats up in September
• Market Review — Gold sinks 1.7%, silver 4% as a 3.7% CPI doubles October hike odds
• TradingView — XAUUSD: Gold falls below $4,150 after US CPI (core 2.6% vs 2.5%, RSI near 31, $4,065-4,070 and $4,020 below, PPI timing)
• FXStreet — Gold braces for US CPI near $4,100 (21/100/50-day SMA stack, RSI, analyst maps)
• Capital Street FX — Gold slips to 4,145 before US CPI (consensus +0.6% m/m, scenario levels, 10-year near 5.22%)
• Capital Street FX — Gold rebounds from 4,066 to 4,194 ahead of CPI (weekly low $4,066.58, payrolls +29k, unemployment 4.2%, central-bank demand ~720t)
• FXEmpire — Bond selloff pressures gold (yield ceiling, support $4,112/$4,073/$4,030, resistance $4,160/$4,190/$4,214)

Related reading: the news board · today's levels · the graded track record.

Risk note: not investment advice. Levels are sourced bands rather than ticks, and spot feeds disagree by $10-20. A CPI print is a repricing event, not a forecast — position sizing, not conviction, is what survives it.