XAUUSD signal: buy the $4,104 retest, out before CPI

Gold closed at $4,138, and it did it against an 84% hike probability

Monday's session opened at $4,056.16 and closed the New York afternoon at $4,138.09 on the desk's Investing.com feed — an $84.08 day range and a 2.0% gain, with the low set at $4,055.14 in the first hour of Asia. Spot was quoted at $4,129.28 when CNBC wrote it up and $4,102.25 at 1125 GMT in Kitco's Reuters copy; the leg into the close took it to $4,138. Treat $4,130-4,139 as the band, not a tick.

What matters is the company that move kept. The dollar index eased 0.37% to 101.84 by mid-afternoon (CNBC) after sitting near 102.17 in the European morning, at the top of its 101.2-102.3 range. The 10-year yield rose to 4.876% and the 2-year to 4.564% (CNBC). CME FedWatch had a December hike at 81% to 84.5% depending on which timestamp you read. Gold rallied into higher real rates — 10-year TIPS near 2.65% — which is not a carry trade. It is a hedge being bought anyway.

Two things were actually sold on Monday, and neither of them was gold. Crude fell after President Trump downplayed fresh strikes on Iran while talks continue, which pulled the inflation premium out of oil and, with it, part of the case for a harder Fed. And the dollar surrendered a slice of Friday's payrolls gain. The PBoC added 25.8 tonnes in October, its largest monthly purchase in three years (CNBC) — a bid that does not check FedWatch before it buys.

That combination is why this is a pullback buy and not a fade. The 4-hour structure has made higher lows since 30 October ($3,870 → $3,919 → $4,029 → $4,055) and Monday closed within $9 of the session high. The medium term disagrees: the 50-day average is still falling at $4,189.73, above price, and October fell 7%. This is a bounce inside a downtrend, not a trend you marry. Session trade, one order, out before the print.

The plan

• Bias: long while $4,055 holds, no shorts into this
• Instrument: XAUUSD spot
• Session: Asia and London, 00:00-13:15 UTC
• Entry: $4,104 buy limit
• Stop loss: $4,042
• Take profit 1: $4,166 — 1.0R, take 50% off
• Take profit 2: $4,228 — 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 $4,150 or below $4,080, or the entry has not filled by 11:00 UTC
• Time stop: flat at 13:15 UTC, fifteen minutes before CPI

Risk is $62, which is 0.76 × ATR(14) of $81.78, measured on the fourteen completed daily bars to Friday. The Sunday stub is excluded from that average deliberately: an 18.97 range on four hours of electronic trading would flatter the stop into something the market eats. The stop sits $13.14 under Monday's low of $4,055.14 — a 0.16 ATR buffer past the structure, comfortably beyond the 0.1 ATR the desk requires. The reference feed for both the low and the stop is Investing's daily and hourly series; exa prints the same low at $4,055.06, and an eight-cent disagreement decides nothing.

Why $4,104

Four consecutive hourly lows between 16:00 and 19:00 UTC — $4,105.27, $4,100.72, $4,100.66, $4,101.44 — and then a $35 extension into the close. That is a shelf, not a guess: the market bought the same $4,100-4,105 strip four times in four hours before it left.

The level also sits on the breakout. The 11:00 UTC hour opened at $4,098.46, closed at $4,108.60, and carried price through Friday's $4,107.50 high. A retest of that candle's body is the cleanest pullback the structure offers, and it is $34 under spot — 0.41 ATR, inside the half-ATR leash.

TP1 at $4,166 is deliberately above the wall. The first real resistance is the 5 November high at $4,147.02, with the 21-22 October shelf at $4,142-4,145 just under it. Any 1R target built on a noise-proof stop lands above those levels — entry within half an ATR of spot plus a 0.7 ATR stop puts TP1 at least $57 above the fill, and the fill is already near $4,104. There is no compliant construction that targets under $4,147. So the trade is: buy the shelf, let it carry through $4,147, bank half at $4,166. If $4,147 rejects, the runner goes to cost and the time stop takes it.

Above that, the 50-day average sits at $4,189.73 and the 15-19 October shelf at $4,177-4,199. TP2 at $4,228 must clear both, inside one Asian and London session, before CPI. Assume it does not.

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 — the outcome the ledger has actually produced most often — the average win is 0.5R and the plan needs 66.7%. That is the desk's known defect, not a fresh discovery: half comes off at 1R and the runner rarely gets paid. Sized against a 66.7% requirement this position is too big. Size it as if TP1 is the whole trade.

Two no-fills in a row

Sunday's signal wanted $4,045 and never got it. Monday's low was $4,055.14 — a miss of $10.14, after which gold rose $83 without the desk in it. That followed 5 November's $4,030 order, which also went unfilled. Two consecutive 0R results that look like discipline on the ledger and are really research paid for nothing: the direction was right twice and the order was a spectator twice. The previous signal is worth reading for exactly how it went wrong.

The fix is not a closer entry for its own sake. It is an entry at a level the market has already traded twice or more in the session being planned — which is what $4,100-4,105 is and what $4,045 was not. If Asia opens above $4,150 and never looks back, this plan voids at 00:00 UTC, and the desk will say that on the track record rather than dress a no-fill up as patience.

What breaks it

• Pre-CPI de-risking. The print lands at 13:30 UTC and the market is long into it. Gold can sag on positioning alone, with no new information at all. Hence no entry after 11:00 UTC and a hard exit at 13:15.
• A dollar reclaim. DXY at 101.84 is 0.6% off Friday's 102.2 close. A push back through 102.3 removes the bid underneath this trade.
• Oil. The Iran de-escalation is the mechanism that let gold rally with hot rates. The Hormuz attacks have not stopped; a reversal reverses the mechanism.
• The 50-day at $4,190. If $4,147 rejects on the first attempt, the falling 50-day becomes a ceiling rather than a target.
• The 09:30 London open. If price runs to $4,140+ without ever trading $4,104, cancel the order — do not chase a fill $30 higher.

Execution notes: XAUUSD spread runs 20-35 cents at a normal Asian open and $1 or worse in the fifteen minutes after 13:30 UTC, which is one more reason not to hold through it. The stop is a level, not a promise — a fast print can fill $3-5 through it. Limits fill at price or better in ordinary liquidity; in the CPI window they do not.

No position through the print. The plan is finished at 13:15 UTC whatever the P&L. The reaction is Tuesday's article, not today's order. The CPI consensus is +0.3% m/m on the headline after September's +0.6%, and +0.2% on the core (Mercury, CNBC); it is the last inflation report before the Fed meets on 9 December, which is the whole reason the levels matter more than the direction here. Background on how the desk reads this hub is on /topics/cpi, every published plan is graded on the track record, and the rest of the week's setups sit on the /signals board.

Sources

• CNBC — Gold climbs more than 1% as inflation fears ease on lower crude prices
• CNBC — Dollar weakens against euro, yen as Iran talks lift risk appetite
• CNBC — Treasury yields rise as investors await October CPI and producer prices
• Mercury — Daily Brief, 9 November 2026
• Kitco/Reuters — Gold gains as lower oil prices ease inflation fears
• Desk data: Investing.com XAUUSD daily and hourly bars, 9 November 2026

Signals are a record of the desk's own reasoning, not financial advice. Levels are quoted for spot XAUUSD on the Investing.com feed; your broker's price, spread and swap differ, and a stop that is structurally correct on one feed can be noise on another.