US core PCE — scenarios before, scorecard after

Before — what we said

August Core PCE m/m is expected at 0.3%, after 0.2% last time. Final Q2 GDP (1.5% expected) and ADP (73K expected) print alongside it, but Core PCE is the Fed's inflation gauge, so it is the main axis. We make a call only on EURUSD, the one asset whose release-day range clearly differs from normal days (1.22x). Gold (1.07x) and BTC (1.04x) behave almost like ordinary days, so we deliberately leave them out. The house's bearish gold and neutral BTC views still stand; this event simply isn't what drives them. The main call is neutral, because the inline branch carries the most weight and the market has already priced a hawkish outcome.

The macro picture right now

The 2y yield at 4.81% (z +2.71) already prices an October hike, so a 0.3% print tells the market nothing new. The 10y real yield sits at 2.83% (z +3.03) while 10y breakevens hold at 2.34% (z +0.4). The rise in yields is therefore real, not a long-run inflation scare, and a hot print would push further on a variable that is already at an extreme. MOVE at 101.8 (z +2.92, up 20.6 points in 5 sessions) shows a jumpy bond market, which amplifies any given surprise. Speculative EUR positioning at -6.4% of OI (z -1.43) means the short side is crowded. So a soft print could move prices more than an equally hot one: a hot number confirms what is priced, while a soft one forces shorts to cover.

How the last print went

The series cooled from 0.44% (January) to 0.15% (June), then ticked up to 0.25% in July. Its 12-period range is 0.15–0.44% with a 0.26% median, so 0.3% is ordinary. Same-period clues lean hot: August CPI and PPI both printed 0.40%. That clue can mislead for two reasons. First, headline CPI carries rising energy prices (WTI +9.6% over a month), which Core PCE strips out. Second, PCE weights differ from CPI. On the same lane, Governor Collins backed hikes and oil broke above $100, but both are already in current yields. Last time (07-30) we leaned bullish on all three: gold was right (+0.735%), EURUSD was right (+0.428%), and BTC went flat (-0.167%). The BTC result is one more reason to drop it this time.

What it reaches, and how

EURUSD has the shortest path: Core PCE → 2y and real yields → the US–euro rate differential → EURUSD. The correlation with US10Y is r=-0.36 ±0.11, so the link is real. Still, no single axis is strong enough to call dominant for EURUSD. Gold sits on the real-yield lane (r=-0.39 vs US10Y), but its current dominant driver is the dollar (r=-0.54, level with VIX). The print has to route through USD before reaching gold, and release-day history shows that detour makes little difference. BTC is driven by the Nasdaq (r=+0.41), so the path runs yields → tech equities → BTC. That is two indirect steps, and the data show little force getting through.

Scenarios when the number lands

Three branches, only ONE happens. Each carries its own numeric threshold and an expected direction per asset.

1. Price moved most like this branch 2 of 3 assets agreed

   Numbered threshold

   Core PCE m/m >= 0.4%

   If this branch plays out:▼gold▼BTC▼EUR30%likelihood

   A hot print confirms the August CPI/PPI trend, and markets begin pricing another hike beyond October. The force runs through real yields that are already extreme. Because most of it is priced, EURUSD more likely drifts in an orderly way toward 1.1356 than breaks sharply.

2. Numbered threshold

   Core PCE m/m = 0.3%

   If this branch plays out:▬gold▬BTC▬EUR45%likelihood

   An inline print leaves October pricing untouched. Within hours, prices return to being driven by forces outside the event: Iran talks, oil, and Treasury supply.

3. Numbered threshold

   Core PCE m/m <= 0.2%

   If this branch plays out:▲gold▬BTC▲EUR25%likelihood

   A soft print contradicts the CPI/PPI clue and puts the October hike in doubt. Crowded EUR shorts plus elevated MOVE could lift EURUSD sharply through 1.1392. Gold could bounce from oversold territory (H4 RSI 23.5), though this event is not usually gold's main driver.

Where we commit — and what would prove us wrong

Unlike the branches above, these carry no “if”. They are the lines graded against real prices below, and they are never edited after the print.

• EUR/USDSIDEWAYSConviction 4/10

  Invalidated if: EURUSD closes an H4 candle outside 1.1356–1.1392 within 4 hours of the release (below 1.1356 tilts bearish, above 1.1392 tilts bullish)

On a soft print, the key question is whether the 2y yield leaves the 4.8% area far enough to erase the October hike. That is the direction current positioning is least hedged for.

Before — what we said

Core PCE m/m lands at 12:30 UTC on 26 August, consensus 0.2% after 0.1% last month. Released alongside it: Prelim GDP q/q at 1.5% (unchanged from prior) and Prelim GDP Price Index q/q at 6.2% (unchanged) — neither carries a built-in surprise, so the reaction concentrates on the core PCE line. We publish calls on EURUSD and gold and deliberately skip Bitcoin: on release days for this indicator BTC's range is 2.90% versus 2.72% on ordinary days, a ratio of 1.07 over 17 prints — indistinguishable from a normal session, and the yield channel into BTC is not measurable. The gold call is carried mainly by the standing bias rather than by the print itself; the in-line branch is the heaviest one and it changes nothing for gold.

The macro picture right now

The 10-year real yield sits at 2.4% with z +2.04 and the 2-year at 4.24% with z +1.73: the real cost of money is at a cycle high and should be weighing on gold — yet the broad dollar is at 118.063 with z -1.62 and gold is still up 8.07% on the week. What is driving gold is the dollar, not yields, so a surprise reaches gold only if it reaches the dollar first. The 10-year breakeven at 2.32%, z +0.03, says the market has no inflation surprise priced in either direction; the miss will therefore show up in nominal and 2-year yields rather than in inflation expectations. EUR speculative positioning at -7.341 %OI, z -1.6, is a crowded short: a cold print gets amplified into a squeeze instead of a drift. The mirror image is gold at 54.691 %OI, z +1.26, plus H4 RSI 80.0 — the long side is already full, so the cold branch lacks fresh fuel and the hot branch has an air pocket underneath. HY at 2.7% (z -1.05) and MOVE at 73.975 (z +0.09) confirm that Nasdaq -3.24% on the week is equity de-grossing, not systemic stress: this print has no detour through the risk-off channel.

How the last print went

The series has been grinding lower: a 0.44% peak in January, then 0.39% → 0.30% → 0.25%, a bounce to 0.33% in May, then a drop to 0.13% in the latest print — one outlier after several orderly months, the kind of number that often gets partly reversed. The 12-period range is 0.13%–0.44%, median 0.25%, standard deviation 0.09; consensus at 0.2% is a mild re-acceleration that still sits below the median, while 0.3% is inside the range rather than a far tail. The strongest clue comes from the two indicators covering the SAME reference month, July: CPI m/m at 0.07% and PPI m/m at -0.03%, both very cold, tilting the balance toward ≤0.2%. Where that clue can fail: part of core PCE is imputed pricing (healthcare, portfolio-management fees) absent from the CPI basket, and we have no fitted coefficient between the two series. There is a counter-catalyst on the lane: a Fed official said further hikes may be needed if inflation does not cool. On 30 July we went 2/3 — gold +0.735% and EURUSD +0.428% with our bullish bias, BTC called bullish but printed -0.167%.

What it reaches, and how

EURUSD has the shortest path: us10y r=-0.34 ±0.12 is a measurable relationship (above the noise floor), and on release days for this indicator EURUSD's range is 1.23 times a normal day — the highest of the three. The chain: print → 2-year and real yields → rate differential → EURUSD, with a crowded short layered on top. EURUSD's currently dominant lane is haven flow (vix r=-0.45), so a hot print severe enough to pull equities down stacks both channels the same way, while a cold print runs on one channel only. Gold has to go the long way round: us10y r=-0.24 ±0.12 is below the noise floor and is therefore not evidence, so the real route is print → yields → DXY (r=-0.55) → gold; the 1.10 release-day ratio says that route is open but narrow. Note the regime is off-textbook: gold–nasdaq +0.34 and gold–vix -0.45, meaning gold currently trades as a liquidity asset, not a haven. Bitcoin has no route from this print: us10y r=-0.11 and wti r=-0.18 are both below the noise floor, it only receives the residual of the dollar (r=-0.46), and it is pinned at 79,005.

Scenarios when the number lands

Three branches, only ONE happens. Each carries its own numeric threshold and an expected direction per asset.

1. Numbered threshold

   Core PCE m/m ≤ 0.1%

   If this branch plays out:▲gold▬BTC▲EUR35%likelihood

   The print matches or undercuts last month's cold reading, consistent with same-month CPI and PPI. The 2-year steps back from z +1.73, the broad dollar takes out another leg and the crowded EUR short gets squeezed — the plausible few-hour travel is the 1.1678 resistance (4 touches) up to 1.1685, roughly one H4 ATR. Gold benefits second-hand through DXY but is held back by its own crowded book, with the nearest resistance at 4,710.1 carrying only 2 touches.

2. Numbered threshold

   Core PCE m/m = 0.2% (in line with consensus)

   If this branch plays out:▬gold▬BTC▲EUR45%likelihood

   No new information for the Fed path: the knee-jerk usually dissolves inside the H4 ATR (38.2 points for gold, 0.0016 for EURUSD) and the narrative reverts to the Treasury buyback programme. EUR still grinds higher as the short side pays to hold against a dollar that cannot rally despite cycle-high yields; gold goes nowhere because its channel needs a hit to the dollar that this branch does not supply.

3. Numbered threshold

   Core PCE m/m ≥ 0.3%

   If this branch plays out:▼gold▬BTC▼EUR20%likelihood

   The print lands above the 12-period median and confirms 0.13% was a one-month artefact. This is the only branch that flips the standing bias: the 2-year extends from an already extreme level, the 'the Fed may have to hike' strand on the lane gets something to hold onto, the dollar is bought back and both assets lose their support at once. For gold the bigger risk is not the size of the drop but the combination of RSI 80 and a crowded long side.

Where we commit — and what would prove us wrong

Unlike the branches above, these carry no “if”. They are the lines graded against real prices below, and they are never edited after the print.

• EUR/USDBULLISHConviction 6/10

  Invalidated if: EURUSD breaks 1.1656 (the 7-touch level) and is still below it four hours after the release.

• Gold (XAU/USD)BULLISHConviction 5/10

  Invalidated if: Gold breaks 4,663.9 (the 10-touch level) and is still below it four hours after the release.

The thing to watch is not the headline but the 2-year: at z +1.73 it has largely finished pricing sticky inflation, so the loose end is the other direction — if the cold reading is confirmed, that is the variable that drags the dollar, not breakevens.

Before — what we said

Four numbers land at 12:30 UTC, but the market really prices two: Core PCE and jobless claims. Advance GDP at 2.1% against 2.0% prior is a snapshot of a quarter already gone; Core PCE decides what is left of Fed hike premium. What makes this release unusual is the backdrop: risk-off is genuine — VIX 20.66, Nasdaq -6.2% on the week, EM equities at 3.5-month lows — yet the dollar will not bid, with DXY at 100.93 and the 30-year at pre-2008 highs. A bear-steepening curve alongside an offered dollar is the signature of fiscal repricing, not of rate fear. That is why gold, BTC and the euro are all supported while equities fall. We keep the running bias; only a clearly hot Core PCE overturns it, because that is when hike premium returns and the dollar finds buyers again.

Scenarios when the number lands

Three branches, only ONE happens. Each carries its own numeric threshold and an expected direction per asset.

1. Hot

   If this branch plays out:▼gold▼BTC▼EUR

   Core PCE m/m at 0.4% or above (or GDP Price Index q/q at 3.0%+), GDP at 2.6%+ and claims below 220k: Fed tightening premium returns, the curve bear-flattens, and the dollar gets its first real reason to bid all week. This is the only path that flips all three running biases at once.

2. In line

   If this branch plays out:▲gold▲BTC▲EUR

   GDP 1.8%–2.4%, Core PCE 0.2%–0.3%, GDP Price Index below 2.7%, claims 210k–255k: nothing forces the Fed to respond and the fiscal channel keeps control. The dollar stays offered, gold and the euro retain their support, and BTC continues to outperform equities. This is the path we consider most likely.

3. If this branch plays out:▲gold▬BTC▲EUR

   Growth miss with sticky prices — GDP at 1.5% or below while GDP Price Index is 2.8%+ or claims are 265k+: the market prices additional cuts, real yields fall, and gold and the euro benefit most clearly. BTC stays neutral as the liquidity bid is offset by equity-driven risk selling.

Where we commit — and what would prove us wrong

Unlike the branches above, these carry no “if”. They are the lines graded against real prices below, and they are never edited after the print.

• Gold (XAU/USD)BULLISHConviction 6/10

• BitcoinBULLISHConviction 5/10

• EUR/USDBULLISHConviction 6/10

The thing worth watching is not the headline but whether the 30-year yield and the dollar move together or apart after the print: steeper with a weaker dollar confirms the fiscal channel, steeper with a stronger dollar means the Fed is being repriced and the current bias breaks.