The call first
December hike odds rose 10 points in a single session. Gold fell 0.47%. This is not the market failing to read the payrolls report. It is gold losing its ability to price the same message twice.
1. The numbers, fixed to a timestamp
Collected 07:21 CST, Nov 7, 2026. It is Saturday: NY gold and Shanghai gold are both marked closed, spot London gold is still quoting, and liquidity is thinner than on a weekday. Every level below is spot XAU.
| Item | Value | Source |
|---|---|---|
| Spot gold | $4,066.11, -$19.38 (-0.47%) | Tonghuashun, 07:21 Nov 7 |
| Session high / low | $4,106.74 / $4,025.87 | Tonghuashun |
| Prior close / open | $4,085.49 / $4,085.68 | Tonghuashun |
| NY close, spot | $4,071.39, -0.4% | Wallstreetcn, 06:15 Nov 7 |
| COMEX gold futures | $4,079.50, -0.83%, +4.63% w/w | NBD, 07:12 Nov 7 |
| Weekly gain, spot | +5.06% | Wallstreetcn, 06:15 Nov 7 |
| Dec 25bp hike odds | 68% Thursday to 77-78% at the close | CME FedWatch, multi-source |
| Spot silver | $51.952, +0.48%, +5.46% w/w | Tonghuashun / NBD |
Two decimal-level discrepancies need flagging up front. The session low is $4,025.87 on Tonghuashun and $4,028.49 on Wallstreetcn. The weekly gain is 4.63% on COMEX futures and 5.06% on spot. Neither is wrong; they are different instruments sampled at different times. Levels here follow Tonghuashun; both weekly figures are given.
2. How much did it actually fall
The data hit at 21:30 CST. By 21:33 spot gold was down 1.15% at $4,041.89 (CLS). From the prior close of $4,085.49 to the low of $4,025.87, the maximum drawdown was $59.62, or 1.46%.
The close was $4,066.11 — a $40.24 bounce off the low, and a net decline of $19.38.
Keep those three numbers apart: 1.46% maximum drawdown, 0.47% closing decline, +4.63% weekly gain. They describe three different things. The drawdown is stop-loss flow. The closing decline is macro repricing. The weekly gain is the actual story of the week. Blend them and you get "payrolls crushed gold," which does not square with the closing price.
3. Ten points of odds, 0.4% of price: four layers
Layer one: the hike news was already paid for.
Into Thursday's close, December hike odds already sat at the 68% handle (Reuters pre-data figure, cited by Fidelity at 13:25 Nov 6). What the payrolls print did was cut the no-hike tail from 32% to 22-23%. It did not build a new expectation from zero; it moved an old one from "probably" to "basically yes."
An expectation that has been on the board for a full month delivers less damage the second time it is confirmed. That is a rule of rates markets, not a quirk of gold.
Layer two: real rates moved only 6bp.
The 10-year yield rose 6.3bp to 4.898% (Wallstreetcn, 21:36). The 2-year rose 5bp (Newsis, 00:36 Nov 7). Read together: the front end moved 5bp, meaning the market did not write a hiking cycle into the curve — it only locked in December. Gold competes with real rates, not with the probability number. A 6bp nominal move with inflation expectations unchanged is a 6bp real move. That maps to a single-digit or low-double-digit dollar adjustment in gold, not $60. The extra $50 inside that drawdown was technical stop flow, not macro.
Layer three: $4,000-4,030 has a bid.
This zone has been tested and bought repeatedly this year, with central-bank and physical demand underneath. That money does not watch CME percentages. It watches price. When price reaches a level it considers cheap, it buys, whether odds read 68% or 78%.
That is why $4,025.87 produced a $40 bounce. It is not a chart line with buyers behind it by accident; it is a level where real bids sit.
Layer four: 102 is a ceiling for the dollar.
TradingKey's headline on Nov 6 put it plainly: December hike odds jump to 77% as the US Dollar reclaims 102.00. Dollar back above 102 and odds at 77% arrived together — and the dollar could not extend much beyond 102, which is exactly why gold did not break $4,000.
4. The other side: when this does break
Four layers explain why it held. The failure conditions deserve equal space.
Condition one: CPI on Nov 10 also comes in hot. September CPI was 3.7% y/y (FRED). If October matches or exceeds it, December odds push from 77-78% toward 85%+. That breaks the "already priced" premise — not an old expectation confirmed, but a new one created. Creating a new expectation hits an order of magnitude harder.
Condition two: $4,025.87 breaks and closes below. That is the week's low and the last line for longs. Once it goes, stop flow returns, and this time there is no $4,000 round number in front to slow it.
Condition three: silver turns down and the gold/silver ratio reclaims 79. Silver gained 5.46% this week and outran gold. If silver starts catching down, the risk upgrades from "rates pressing gold" to "systematic deleveraging," and gold's decline stops being measured in tens of dollars.
Two of these three together void the "it held" call.
5. The weekly chart is the real story
Chain the week together. Nov 4: COMEX gold +3.34%, from a $3,928.2 low to $4,072.2. Nov 5: pushed to $4,146.99, faded, closed $4,087.15. Nov 6: payrolls drove it to $4,025.87, closed $4,066.11.
The weekly candle is a long-legged doji-ish bullish bar, +4.63% on COMEX. The lower wick sits precisely at $4,000-4,030, which matches layer three: that zone absorbed $60 of selling inside a single session.
On the daily it reads as "payrolls smashed gold." On the weekly it reads as "pullback and hold." Same facts, two timeframes, two conclusions. Short-term traders read the first, trend traders read the second. Both are defensible; mixing them is what produces bad calls.
6. Silver as corroboration
Spot silver $51.952, still +0.48% on the day and +5.46% on the week, beating gold by 0.8 percentage points. The gold/silver ratio compressed from 79.4 on Nov 5 to 78.26.
The payrolls minute is the tell: gold -0.82%, silver +0.66% at 21:40. Metals that refuse to sell off together in a risk event are telling you what kind of sell-off this is — not a liquidity squeeze, but the rate side pressing gold alone while silver's industrial leg holds it up.
7. Closing thought
Yesterday I published a piece saying hike odds "moved only 0.2pp." That was a 21:37 intraday print, written three hours before the story ended. By the US close the number was 77-78%. I have corrected the headline and the body; the error stays visible in the text.
That episode is worth recording: on a data day, an intraday print is three hours away from being a different story. Writers and traders share the same failure mode — concluding inside the first candle.
Back to today: the market is not refusing to accept a hike. It is refusing to pay for the same message twice. The first payment was in early October, when odds went from 30% to 68%. The second payment needs a new reason, and CPI is the next candidate.
One question to leave with: if CPI on Nov 10 comes in soft and the $4,085.49 gap gets filled, does that mean the 10-point repricing has already been absorbed?
Data collected 07:21 CST, Nov 7, 2026. Quotes: Tonghuashun precious metals (spot gold, spot silver, NY gold); closing and weekly figures from NBD 07:12 and Wallstreetcn 06:15. Hike odds: CME FedWatch, cross-checked across Reuters (68% pre-data), Sina 21:42 (72.8%), TradingKey / Invezz / AInvest (77%), Guanghuan Market Broadcast (78%); expressed as a range. Event data: BLS October employment report, 21:30 CST Nov 6. CPI prior: 3.7% y/y for September, FRED.
Structural analysis, not investment advice. No return is promised.
Related
Check out more trading insights in the Trading Journal.