XAUUSD signal: sell the $4,048 bounce, target $4,000

Gold settled Thursday at $4,022.34 on Investing.com's XAU/USD series, down $45.32 or 1.11% from Wednesday's $4,067.66 close. A second feed had it at $4,020.56 at the same moment, so call the band $4,020–$4,023. The session opened $4,054.99, printed its high at $4,068.72 in the 01:00 UTC bar and its low at $4,005.38 eleven hours later in New York.

It never traded above its opening price all session. The interesting part is what else fell.

The bounce died at the 61.8% retrace

The 22 October collapse ran from $4,145.40 to $3,975.64, $169.76 in two sessions. Wednesday's FOMC spike topped at $4,082.79 — $2.24 above the 61.8% retracement of that leg, at $4,080.55 — and stalled there. Thursday's high, $4,068.72, was a lower high, and the close was $46 under it.

The recovery had a cleaner tell than the highs: its lows.

• 23 October: $3,975.64
• 26 October: $3,978.15
• 27 October: $3,988.14
• 28 October: $4,024.36
• 29 October: $4,005.38 — the ladder ends here

Four rising lows is a base. A lower low is a range with a broken rung, and Thursday broke it. $4,060 is where the 22 October breakdown started. The reclaim test needed two closes above it; gold delivered one — Wednesday's $4,067.66 — and handed it back on Thursday. The close is $38 beneath it.

The 20-day average sits at $4,121, recomputed on completed bars after Thursday's close, and price is 2.4% below it. Until gold closes above that line, a rally and a bounce are the same thing.

Gold fell while the dollar and the yield fell too

Here is the part that does not fit the standard explanation. Both of gold's usual drivers were bid on Thursday:

• The dollar fell. The index closed at 101.9753, down 0.21% (Trading Economics).
• Yields fell. The 2-year dropped 3bp to 4.693% and the 10-year 3bp to 5.011% after the inflation print, with the 10-year quoted near 4.98% late in the session (CNBC, Trading Economics).

A weaker dollar and a lower 10-year, and gold still lost 1.1%. That is not a rate story. It is a seller who needed liquidity.

The data explains none of it, because there was nothing in it. Core PCE matched consensus at 0.3% m/m and 3.1% y/y; headline printed 0.5% m/m and 3.6% y/y, a tenth under the 3.7% estimate; Q3 GDP came in at 3.4% against 3.1%–3.2% expected (CNBC). December hike pricing did not move: 83% before the print, 83% after it.

The selling came earlier. Between the 01:00 bar and the 08:00 bar gold slid from $4,068.72 to a $4,016.02 low — $52.70 across Asia and the London open, four and a half hours before the BEA published anything. The desk's own write-up of the print, core PCE in line, measured the one-hour reaction at $0.52. There was nothing there to trade.

What is left is energy and Europe. WTI settled Wednesday at $102.78, up 2.9%, and Brent at $115.32 on reports of a strike on a Saudi refinery (FXStreet). The ECB held its deposit rate at 2.5% with eurozone inflation at a three-year high, hours after Christine Lagarde said she would leave in February with French bonds unsettled (CNBC, Euronews).

$115 crude, a Fed with two members voting to hike and a euro with a leadership vacancy all point real-yield risk the wrong way for a metal that pays nothing. The war bid and the rate bid used to be one trade. This month they are not.

Why $4,048 is the level now

Wednesday's plan parked a buy limit at $4,048 and it filled on Thursday's 00:00 UTC bar. The best price after the fill was $4,068.72, $34 short of the $4,103 target, and the $4,005.38 low never reached $3,993. The plan's own invalidation ended it: two hourly closes under $4,040 at 03:00 ($4,038.25) and 04:00 ($4,038.12). Graded there, the fill lost $9.88 of $55 — −0.18R, not the −0.52R of holding to the 11:30 time stop.

The level outlives the trade that used it, in the other direction. $4,048 is the top of Thursday's Asian breakdown: the 05:00 UTC bar printed $4,046.40 and then the market made lower highs for fifteen hours — $4,037.70, $4,039.49, $4,037.61, $4,027.83, $4,023.22. Broken support that has not been reclaimed is the cheapest place to sell a bounce, and $4,048 is where a bounce dies if one comes at all.

The plan

• Instrument: XAUUSD spot, MT5 gold
• Session: Friday 30 October, 00:00–20:00 UTC — 08:00 Friday to 04:00 Saturday SGT
• Bias: lower while $4,068 caps it; a close above $4,082.79 ends the sequence
• Entry: $4,048 sell limit
• Stop loss: $4,096
• Take profit 1: $4,000 — 1.0R, close half
• Take profit 2: $3,952 — 2.0R, close the rest
• Risk-reward: 1:1 to TP1, 1:2 to TP2 on the full plan
• Do not take it if: gold prints an hourly close above $4,068 before the order fills. The shelf would be reclaimed and the order becomes a fade of a breakout.
• Time stop: flat or unfilled by 20:00 UTC Friday

The arithmetic, so it can be checked rather than trusted:

• Stop distance $48 = 0.725 × ATR(14) $66.20, computed on 14 completed daily bars. The desk's floor is 0.70 × ATR, or $46.34. A tighter stop would sit inside Thursday's $63.34 range — the noise band that decided this month's losses.
• Entry is 0.39 × ATR above spot ($25.66), inside the half-ATR limit that keeps a resting order honest. A zone further away looks disciplined and never trades.
• The stop sits $13.21 above Wednesday's $4,082.79 high — a 0.20 × ATR buffer beyond the structural level, priced off Investing.com's daily series. The buffer exists because a $5 feed disagreement once decided a verdict here.
• Break-even win rate: 40% if both targets print (0.5R at TP1 plus 1.0R average on the runner, against a 1R loss) and 66.7% if the runner is scratched at entry. A 1R first target against a noise-aware stop needs 50%. This one needs 40%, and TP1 is $5.38 below Thursday's low instead of $100 away.

What voids it

The plan is wrong, not early, on any of these:

• An hourly close above $4,068.72. Thursday's high and the top of the shelf. One close is enough to say the sellers have gone.
• A close above $4,082.79. Wednesday's high and the 61.8% retrace. Above it the lower-high sequence this plan is built on stops existing, and the next reference is the 20-day at $4,121.
• No fill. If Friday stays under $4,040 all session there is no trade. Do not chase it lower at market; in a range this wide the edge is the level, not the direction.
• Downside levels: an hourly close below $4,005.38 opens $3,988.14 (Tuesday's low) and then the $3,975.64–$3,978.15 floor that has held five sessions.

TP2 at $3,952 needs that floor to break and extend — a range resolution, not an intraday flush. That is the honest stretch: TP1 is a $6 break of Thursday's low, TP2 a fresh five-week low. If the floor holds, the trade pays half a plan.

Friday's calendar, month end and the weekend

• 12:30 UTC — Q3 Employment Cost Index. Consensus 0.8% against 0.9% in Q2, and the last October release that speaks to the wage side of a 3.6% inflation print (BLS schedule).
• 13:45 UTC — Chicago PMI, 58.8 prior. Second-tier, but it lands in the London-New York overlap where this month's damage has been done.
• Month end. 31 October is a Saturday, so Friday carries the rebalancing fix — the one flow that can push gold through a stop for no informational reason, which is why the stop sits $13 above Wednesday's high rather than $2.
• The weekend. The desk's study of 17 Sunday-reopen gaps from July to October put the median at $4.52 and the worst at $39.45 against an ATR of $66.99. No gap in that sample came close to a 0.7 × ATR stop, so flattening before the close is about not carrying risk into Tuesday's midterms, not about gap protection.

What to take away

• The bounce that began on 23 October is over: four rising lows became a lower low, and the 61.8% retrace held to $2.24.
• Gold fell 1.1% while the dollar lost 0.21% and the 10-year lost 3bp. That is liquidation, and liquidation does not wait for the data.
• The trade is a sell of the $4,048 shelf, $48 at risk, 1R at $4,000 and 2R at $3,952. Break-even is 40%.
• Above $4,068 there is no trade; above $4,082.79 there is no thesis. The rest of the week's level work is on the signals board and under the dollar.

Sources

• CNBC, PCE inflation report, September 2026, 29 October 2026
• CNBC, Treasury yields after the PCE and GDP data, 29 October 2026
• Trading Economics, United States dollar index, 29 October 2026
• Trading Economics, US 10-year government bond yield, 29 October 2026
• CNBC, ECB holds rates at 2.5%, 29 October 2026
• Euronews, ECB holds as Lagarde announces an early exit, 29 October 2026
• FXStreet, Fed holds steady; markets await PCE, GDP and earnings, 29 October 2026
• BLS, Schedule of selected releases, October 2026
• Price data: Investing.com XAU/USD daily and hourly series, pulled 29 October 2026, 20:05 UTC

Not financial advice. Levels and the plan are the desk's reading of one session, not a prediction, and gold can gap through any of them.