Gold pinned at the daily pivot with CPI four hours out

Spot is sitting almost exactly on its daily pivot after a $246 intraday range, which is the polite way of saying somebody got carried out before London open. The daily structure remains constructive — price is above the EMA20 and EMA200 — but every short timeframe has rolled under its moving averages, and the dollar's hourly is overbought into a 12:30 UTC CPI print. The setup is binary: a soft core CPI hands gold the 4,800 weekly R1, a hot print sends it back to the lower half of yesterday's range. Conviction percentages below are qualitative, not back-tested.

The session

Gold is trading 4,707.80 — up 3.62% on the day but coming off a session high of 4,783.40, which puts price roughly $76 below the high and $171 above the low. That's a 5.4% intraday range, which is not a tape you fade with size. The move up overnight ran into supply right where FXStreet's desk tags the 61.8% retracement (4,742) and our weekly R1 (4,800) — gold poked above 4,742 toward 4,783 and got rejected.

Two things drove the rally and the failure. The dollar is bid on the 1h (DXY RSI 69.6, above all hourly EMAs) but bleeding on the day (-0.25%, daily DXY below EMA20/50/200) — the textbook signature of a corrective bounce inside a bigger downtrend. Meanwhile the 10y added 6bp to 4.41%, and rising nominal yields are normally a headwind, so the fact gold rallied through it tells you the bid is coming from somewhere other than the rates curve. FXStreet flags Trump's rejection of Iran's peace proposal as the kicker; that's consistent with the geopolitical risk premium showing up in metals while the dollar refuses to fully cooperate.

The 30-day DXY/XAU return correlation is -0.71, which is strong but not absolute. A hot CPI today can flip that for a session.

Multi-timeframe read

The four timeframes are not telling the same story, which is exactly what you'd expect for a pre-event tape.

• 15m: RSI 33.6, MACD histogram negative but turning up, price below EMA20/50/200. Classic pullback inside a larger move — short-term oversold without confirmation.
• 1h: RSI 43.9, MACD histogram still negative and falling (-4.86), price below EMA20 and EMA50. The hourly impulse from the 4,537 low has lost momentum and price is digesting under the EMAs.
• 4h: RSI 56.2, price below EMA20. The four-hour is the neutral timeframe — slightly bullish RSI, slightly below trend. This is where confluence will resolve once CPI prints.
• 1d: RSI 51.0 (mid-band), MACD histogram +13.93 and rising, price ABOVE EMA20 and EMA200 but BELOW EMA50. The daily trend is intact and momentum is improving on the daily even as the intraday looks heavy.

Where they agree: nobody is at an extreme. Where they diverge: short timeframes are mechanically weak, the daily is mechanically strengthening. That divergence is what produces breakout failure or breakout confirmation depending on which leg gives way. The honest read is that the 4h is the swing vote, and CPI is what hands it the gavel.

Macro frame

DXY at 98.27 sits below its daily EMA20, EMA50, and EMA200 — that's a structural dollar downtrend on the timeframe traders actually allocate from. The 1h overbought reading is a counter-trend bounce, not a regime change. If CPI prints in line or soft, the daily DXY structure is the one that controls.

10y at 4.41% with a 6bp daily move higher is uncomfortable for a metal that pays no coupon. The fact gold still rallied tells you the bid is real-yield insensitive at the margin — i.e., it's haven flow and central-bank-adjacent demand, not a rate-cut trade. ING's framing via FXStreet — "Gold's path higher hinges on Fed easing" — is the medium-term gating factor, but the intraday gating factor is the inflation print and how it moves the front of the curve.

The PAXG/USDT cross on Binance is trading at a 0.30% discount to spot XAU. That's a small but real signal: the 24/7 token tape is not chasing the paper rally. Crypto-side gold buyers are saying "we'll wait for confirmation," which lines up with the hourly RSI/MACD weakness.

COT colour: managed money is net long 163,303 contracts (longs 211,814 / shorts 48,511). That's a sizeable net long but not at the extremes seen at prior cyclical tops. Commercials are net short 198,935, which is the mirror image and what you always see when MM has built a long position. The thing to watch is whether next Friday's COT shows MM longs adding through 4,800 or capitulating below 4,650 — both moves would tell you the post-CPI direction stuck.

Two scenarios

These are qualitative conviction estimates, not probabilities derived from a back-test.

Buy setup

• Trigger: 1h close back above the daily pivot at 4,722.30, with a follow-through 15m printing above the session VWAP. Bonus confirmation if it coincides with a soft Core CPI MoM (≤0.2% consensus) at 12:30 UTC.
• Invalidation: 1h close below 4,696 (FXStreet's 50.0% Fibonacci) — that level lines up with weekly P at 4,656 as the next floor, so anything that loses 4,696 is no longer a long.
• Target: First scale 4,742 (61.8% retracement and session resistance shelf). Runner to 4,800 (weekly R1). A clean break of 4,800 opens the daily R-cluster at 4,879 (weekly R2).
• Conviction: 55%
• Rationale: Daily trend is intact, daily MACD is turning positive, DXY structure is bearish on the daily, and managed money is positioned the right way for continuation. The 15m being oversold into a known catalyst is the kind of pullback you want to be buying if the catalyst cooperates. The wrinkle is the hourly DXY overbought — a hot CPI can break this trade in twenty minutes.

Sell setup

• Trigger: 1h close below 4,696, OR a knee-jerk spike toward 4,742–4,750 on a hot CPI that fails to hold above 4,725.90 (daily R1) on the next 15m candle. The trade is selling strength into a hawkish surprise, not chasing weakness.
• Invalidation: 1h close back above 4,742. That reclaims the FXStreet 61.8% level and the daily R-cluster — at that point the short premise is gone and the buyers are paying up.
• Target: First scale 4,656 (weekly pivot). Runner to 4,576 (weekly S1), with knowledge that S1 lines up with the lower half of the recent five-session range (4,663-4,783 base).
• Conviction: 35%
• Rationale: Daily structure is the wrong side for size shorts, but the intraday tape is heavy enough that a hot CPI + DXY 1h overbought reflex could produce a fast 80-100 dollar move down before the daily bid returns. This is a tactical short, not a swing — work it small, take the first target seriously, leave the runner to confluence.

Levels worth marking

Sources of confluence pulled from the snapshot pivots and the FXStreet desk note:

• 4,800.17 — weekly R1. The line in the sand for the bull case continuation.
• 4,742 — FXStreet 61.8% Fibo / today's failure point / first scale for longs.
• 4,733.10 / 4,725.90 — daily R2 / R1 cluster. Very tight (prior day HLC was a doji), so this is a thin shelf, not a wall.
• 4,722.30 — daily pivot. Current price action is pinned here.
• 4,718.70 — prior daily low/close (the doji body). A magnet.
• 4,711.50 / 4,704.30 — daily S2 / S3. Bears want these to flush.
• 4,696 — FXStreet 50% Fibo and the buy-side invalidation. Critical.
• 4,656.43 — weekly pivot. The bigger picture floor for the week.
• 4,576.67 — weekly S1, and the area where the recent five-session range bottoms out.

EMA confluence on the daily: price above EMA200 (long-term trend bullish), above EMA20 (short-term trend bullish), below EMA50 (medium-term still working off the prior leg). A daily close back above the EMA50 is the swing-trade signal; we don't have it yet.

Calendar / catalysts

(Per TradingEconomics; impact subjective.)

• Tue 12 May 12:30 UTC — US CPI. Consensus Core MoM 0.2%, Headline MoM 0.5%, Headline YoY 3.6%. The single highest-impact print on the week for gold. Hot = dollar reflex up, gold reflex down. Soft = the buy setup activates.
• Tue 12 May 17:00 UTC — 10y Treasury auction. Tail or stop-through will move the back end of the curve and feed back into the metal indirectly.
• Tue 12 May 18:00 UTC — Monthly Budget Statement. Low immediate impact, but the deficit print is a slow-burn gold tailwind.
• Wed 13 May 12:30 UTC — US PPI + Retail Sales + Jobless Claims. PPI is the inflation follow-up; Retail Sales is the consumer read. Cluster risk.
• Wed 13 May 17:00 UTC — 30y bond auction. Watch for the long-end tail.

Fed speakers aren't on the snapshot list — assume blackout-light, not blackout. Any wire-flash hawkish comment is a fade for gold longs.

Sources cited

• https://www.fxstreet.com/markets/commodities/metals/gold — technical levels (4,742 / 4,807 / 4,890 resistance, 4,696 / 4,671 / 4,651 support), Iran/Trump fundamental colour, ING quote on Fed easing.
• https://tradingeconomics.com/calendar — US CPI / PPI / Retail Sales schedule and consensus.
• (Reuters Markets and Al Jazeera attempted, did not return content this session.)

(not financial advice)