XAUUSD signal: sell the $4,033 retest, $3,967 target

October CPI took $111 out of gold and broke the six-day higher-low structure. The desk wants a sell at the $4,033 shelf, a $66 stop above the pre-CPI pivot, half off at $3,967, and nothing held into a thin Veterans Day afternoon.

Levels in this article
  • $4,138.88
  • $4,141.36
  • $4,013.72
  • $4,027.87
  • $4,139.84
  • $4,023.26
  • $4,008.59
  • $4,023

Trade plan at a glance

Bias
short while $4,099 holds, no longs into this
Entry
$4,033 sell limit
Stop loss
$4,099
Take profit 1
$3,967 — 1.0R, take 50% off
Take profit 2
$3,901 — 2.0R
Risk-reward
1:1 to TP1, 1:2 to TP2, blended 1.5R if both fill
Do not take it if
the 00:00 UTC open is above $4,072 or below $3,990, or the limit has not filled by 11:00 UTC

Primary levels from this article — any conditional alternative is laid out in the body below. Sizing comes from the stop distance; confirm on your own feed before acting, spreads and gaps are real.

Gold gave back $111 of a six-day rally in one CPI print

Tuesday opened at $4,138.88 on the desk's Investing.com feed, printed its high at $4,141.36 inside the first hour, and spent the rest of the session going one way. The low was $4,013.72 at 16:45 UTC and the 20:00 UTC hour was trading $4,027.87 — a range of $127.64 and a 2.7% decline from Monday's close of $4,139.84. A second series, Binance's PAXG, printed $4,023.26 at the same minute and a low of $4,008.59, so treat the close as $4,023-4,028 and the low as $4,009-4,014 rather than a tick.

CNBC's wrap caught spot at $4,051.50, down 2.1%, and December futures 1.3% lower at $4,059.30 — both from before the last $25 of the slide. Silver fell 2.9% to $51.86. Marex's Edward Meir gave the day its one honest sentence: "Both the overall number CPI and the core CPI came in a shade above estimates."

What the print actually did

  • Headline +0.4% m/m against +0.3% expected, from +0.6% in September (investingLive).
  • Core +0.3% m/m against +0.2% expected, unrounded +0.297%, the third consecutive 0.3% month.
  • No annual rate was published — headline or core. The BLS did not collect October 2025 prices during the shutdown, so there is no base index to divide by, and the y/y columns read "not published" against priors of 3.7% and 2.6%.
  • Supercore roughly +0.4%, just over half of it from services less energy.
  • The 2-year went to 4.727%, +18bp on the day, off a 4.547% close (CNBC).
  • The 10-year printed 4.983%, +12.1bp, having touched 5.021% (CNBC).
  • DXY 102.245, +0.58%, with the day's high at 102.41 (CNBC).

CME FedWatch now has a December 25bp hike fully priced and a 50bp move at 13.6%, from 0% before the data; investingLive's read of the curve is 60bp of hiking priced over the next year. That is the mechanism behind a $111 day, and it is not one that reverses on a Wednesday with no data in it.

Why this is a short and not a fade

The desk does not fade a trend, and the last four losing signals all did. Direction here is settled by structure, not by what feels overdue after a $111 fall.

  • The rally made a double top. $4,147.02 on 5 November, $4,148.95 on 9 November, and $4,141.36 today — the third high is the lower one.
  • It broke the higher-low sequence. The ladder ran $3,870.17 (30 October) → $3,919.65 (4 November) → $4,029.95 (6 November) → $4,054.04 (9 November). Today printed $4,013.72 and closed under all but the first two of those.
  • It is still under a falling 50-day. The 50-day sits at $4,185.38 and has dropped $40.93 in six sessions; the 100-day is $4,200.73. Monday's high stopped $36 short of it. That makes the whole six-day, 6.8% advance a counter-trend bounce inside a falling market, and today the bounce failed.

So the trade is a continuation short on the failure of that bounce. What would make it a fade is selling into new highs with the medium-term trend up; price is $157 below its 50-day with lower highs printed on the daily.

The plan

  • Bias: short while $4,099 holds, no longs into this
  • Instrument: XAUUSD spot
  • Session: Asia and London, 00:00-13:15 UTC
  • Entry: $4,033 sell limit
  • Stop loss: $4,099
  • Take profit 1: $3,967 — 1.0R, take 50% off
  • Take profit 2: $3,901 — 2.0R
  • Risk-reward: 1:1 to TP1, 1:2 to TP2, blended 1.5R if both fill
  • Do not take it if: the 00:00 UTC open is above $4,072 or below $3,990, or the limit has not filled by 11:00 UTC
  • Time stop: flat at 13:15 UTC, before the New York equity open

Risk is $66, which is 0.71 × ATR(14) of $93.62. That ATR is measured with today's bar included, on purpose: the fourteen completed bars to Monday average $87.21, and using that older figure would put the floor at $61 and flatter the stop on the day volatility expanded. Size it against the larger number.

The stop sits $19.26 above the pre-CPI pivot at $4,079.74 — the 12:00 and 13:00 UTC hourly highs, where price based for two hours before the release — which is 0.21 ATR of clearance, comfortably past the 0.1 ATR buffer the desk requires. Reference feed for the pivot, the low and the stop is the Investing.com hourly and daily series.

Why $4,033

Four touches between 18:30 and 19:45 UTC: $4,033.38, $4,033.38, $4,032.43, $4,032.40. That is the top of the shelf the market built after the CPI slide stopped, and it is $5.13 above spot — 0.05 ATR, which is the point.

The desk has now had three orders in a fortnight that never filled while the market went the right way. $4,045 missed by $10.14 on 9 November; $4,030 missed entirely on 5 November. Those were parked 0.3-0.4 ATR from spot and the market opened at its extreme and left without them: Monday opened at $4,056.16, printed its $4,055.14 low in the first minute, and rallied $84; today opened at $4,138.88 and printed its $4,141.36 high inside the first hour, then fell all day. Twice in two sessions the open was the extreme. A retest order $30 above spot is a spectator in that market; $5 above is not.

  • TP1 at $3,967 sits just under the $3,970.45 high of 2 November, at the bottom of the $3,970-3,988 shelf that broke on 22 October. It is not a gift: price has to clear the 27 October low at $3,988.14 and the 23 October low at $3,975.64 first, and if the shelf holds, the runner goes to cost.
  • TP2 at $3,901 needs the 4 November low at $3,919.65 taken out and sits above the 3 November low at $3,893.94. Treat that as unlikely inside thirteen hours.

The arithmetic, stated plainly

If both targets fill, the average win is 1.5R and the plan breaks even at a 40% strike rate. If TP1 fills and the runner is scratched at cost — which is what the ledger has mostly produced — the average win is 0.5R and the plan needs 66.7%. The book's measured average win is 0.82R over eighteen graded signals, so size this as though TP1 is the whole trade and the stop is the whole risk.

Yesterday's plan filled and stopped

This morning's article wanted a $4,104 buy limit with a $4,042 stop, and the market delivered both in the same session. The 09:00 UTC hourly bar traded from $4,113.13 down to $4,081.32, so the limit filled at about 09:05. The CPI bar then printed $4,040.46 and took the stop.

Anyone who honoured the plan's 13:15 UTC time stop was out near $4,072 — the 13:15 bar closed at $4,072.41 — for roughly 0.5R, and the stop was hit fifteen minutes later without them. That is the one part of the plan that worked. The rest was a long into a hot inflation print: the reaction article said before the print that a hot number was the risk, and the order was live into it anyway.

What breaks it

  • A thin tape. Wednesday is Veterans Day and SIFMA has confirmed a full close of the US bond market, with cash Treasuries reopening Thursday (SIFMA). No data is scheduled; the next releases are jobless claims Thursday and PPI Friday. With the cash curve shut, dollar moves come off futures only and spreads widen, so a $66 stop can be clipped by a thin print that means nothing. This is a thirteen-hour trade for that reason.
  • A dollar reversal. DXY at 102.245 is 0.6% off Monday's 101.653 close and back at the top of its range. A failure back under 101.5 removes the mechanism.
  • A reclaim of $4,054. Monday's low is the first level that would say the down-break failed rather than paused; the stop at $4,099 is deliberately above both it and the $4,079.74 pivot.
  • The 09:30 London open. If price runs to $4,000 without trading $4,033, cancel — do not chase a fill $30 lower.

Execution notes: XAUUSD spread runs 20-35 cents in a normal Asian open, wider on a US holiday when the interbank market is thin. The stop is a level, not a promise — a fast print can fill $3-5 through it. If the limit fills and price stalls between $4,020 and $4,033 for two hours, take it off.

No position is held into the New York afternoon. With the bond market closed and no data, there is nothing to be early for. Every published plan is graded on the track record, the day's levels are collected at /levels, and the rest of the week's setups sit on the /signals board. Background on how the desk reads inflation prints is on /topics/cpi.

Sources

Not financial advice. Levels are the desk's own read of the Investing.com XAU/USD series at 20:00 UTC on 10 November 2026 and are published before the session they cover; a gap through the stop is a gap through the stop.

Trading these levels? XM — MT5, gold, demo first paid link