US ISM manufacturing PMI — scenarios before, scorecard after

Before — what we said

ISM Manufacturing PMI lands at 14:00 UTC on 1 September, consensus 55.2 after 55.6. Released alongside it: ISM Prices at 71.2 (prior 71.1) and JOLTS Job Openings at 7.33M (prior 7.36M) — three separate numbers, and this time the sub-line matters more than the headline. A 0.4-point gap between forecast and prior is noise; Prices is the only line in this print that touches inflation expectations directly. We publish calls on all three assets because none has a ~1.0 profile: release-day range versus a normal day is 1.35x for BTC (n=19), 1.19x for gold, 1.14x for EURUSD. But the BTC call is a RANGE call, not a directional one — its two channels pull against each other (see transmission). Gold and EURUSD keep the direction of the running view; only the weak-print branch overturns it.

The macro picture right now

The surprise will land on a front-end real-rate shock, not on fear. Ten-year real yields at 2.34% (z +1.68) and two-year yields at 4.20% (z +1.53) both sit at one-year extremes: a strong ISM has very little room left to push through the same channel, while a weak one has plenty of room to give back — an asymmetry that favours the cold branch even though inline remains the heavier one. Ten-year breakevens at 2.31% (-0.02 on the session) are still slipping, so a hot Prices line has to argue with that curve rather than simply add to it. The risk lane is open, not defensive: VIX 14.51 (-4.6%), HY spreads 2.63% (z -1.44) still tightening, Nasdaq +1.43% — a bad number has somewhere to do damage. Dollar strength is narrow: broad USD 118.063 (z -1.62) while DXY is +0.8% on the week. Positioning decides who pays: gold speculative length at 56.859% OI (z +1.59, +8.887 over five sessions) is the most crowded side at the table, with price -5.46% on the week and RSI at 28.3; EUR is already short at -4.441% OI (z -1.24), so most of the bad EUR news is pre-owned; BTC just flushed -6.105 points of OI over five sessions with DVOL at 37.08 (z -0.92) — clean leverage, cheap vol.

How the last print went

We have no free series source for the ISM family, so we cannot measure the indicator's own momentum — that is something we do not know, not something that does not exist. The nearest clue on the same lane and the same August reference month is Chicago PMI at 47.1 against a 57.9 forecast: a miss of more than 10 points, enough to fatten the weak branch relative to a normal distribution around consensus. That clue can be wrong in an obvious way: Chicago is a single-region survey with a small sample, it regularly diverges from the national ISM, and a few large manufacturers can dominate it — it tilts the balance, it does not set the level. On the last scored print (3 August) we went 0/2: we called gold lower and it moved 0.1%; we called EURUSD higher and it moved -0.15%. Both stood still. The lesson is dosage rather than direction — we overestimated this release's ability to move price on its own. Hence confidence caps at 5 here, even with a clearly bearish base view.

What it reaches, and how

The event runs on the Risk-appetite lane. For BTC that is almost its home lane: nasdaq r=+0.36 ±0.11, a direct path. But BTC's dominant lane is Dollar strength at r=-0.45, level with vix inside the error band — so a strong print lifts BTC through the risk leg and presses it through the dollar leg, leaving a small net move while the day's range still widens. That is why the BTC call is a range call, not a direction. For gold the path has to detour: the nasdaq coefficient r=+0.33 ±0.12 carries the sign OPPOSITE to theory (gold is trading WITH equities), so we do not lean on it; the real chain is ISM/Prices -> breakevens and the front end -> real yields -> DXY -> gold, with DXY the measured dominant channel at r=-0.55. EURUSD has two parallel paths: directly through the front-end rate gap, and indirectly through its dominant Haven/fear lane (vix r=-0.44) — the two align on a strong print and conflict on a weak one.

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

   ISM Manufacturing PMI >= 56.0 OR ISM Prices >= 73.0 (consensus 55.2 and 71.2)

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

   The Prices line clearing its threshold stops breakevens slipping while the front end is reinforced: real yields extend from an already extreme level, DXY benefits, and the crowded gold longs are the side that pays. A reasonable few-hour distance is still under one H4 ATR (gold 41.8), because most of the tightening narrative is already in the price; EURUSD travels the same way but slower, since positioning is short before the fact.

2. Numbered threshold

   PMI within 54.7–55.7 · Prices within 70.0–72.5 · JOLTS within 7.20M–7.50M

   If this branch plays out:▼gold▬BTC▼EUR50%likelihood

   No new information for the curve, so price returns to what is already driving it: the front-end rate gap and a crowded gold long that needs an exit. Gold is likelier to leak toward the 4,405.2 area (6 touches) than to bounce; EURUSD holds direction but moves slowly, pinned against 1,1586 (8 touches) and 1.1586 (9 touches) where profit-taking by the short side is the main resistance; BTC stays between its two boundary levels.

3. Numbered threshold

   PMI <= 54.0 OR JOLTS <= 7.15M (prior 55.6 and 7.36M)

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

   This is the only branch that overturns the gold call: a growth shortfall -> less conviction in the tightening path -> the front end gives back -> the dollar loses its support, and gold gets both its rate leg and its dollar leg. For EURUSD the branch argues with itself — a narrowing rate gap lifts it, but its dominant lane is fear and a US growth miss usually lifts VIX — so the net is close to zero. BTC takes the event's own lane straight on.

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)BEARISHConviction 5/10

  Invalidated if: Gold trades above 4,454.2 (9-touch resistance) in the hours after the release

• EUR/USDBEARISHConviction 5/10

  Invalidated if: EURUSD trades above 1.1610 (5-touch resistance) in the hours after the release

• BitcoinSIDEWAYSConviction 4/10

  Invalidated if: BTC closes the first H4 candle after the release outside 76,737–78,160

Read the Prices line before the headline: it decides whether a strong ISM gets read as 'growth is fine' or as 'the Fed still has work to do' — two readings that send gold and EURUSD in different directions off the same PMI number.

Before — what we said

This is the week's first US print, landing Monday at 14:00 UTC. The backdrop is unusual: US10Y at 4.745% and up 6.03% on the month, a hike expected at the coming FOMC, and yet DXY closed at 99.80 after a 1.65% weekly decline. The dollar is refusing to be paid for its rate advantage — which is why EURUSD keeps a bullish bias. We are making calls on gold and EURUSD only. Bitcoin is deliberately left out: the transmission from a manufacturing survey to BTC barely exists, and its standing bearish bias comes from Strategy's authorisation to sell up to $5bn of Bitcoin — a coin-market matter, not an ISM one. Note that gold, DXY and EURUSD here are Friday closes, with a Sunday 21:00 UTC gap still ahead of them.

Scenarios when the number lands

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

1. Hot

   Numbered threshold

   ISM Manufacturing PMI ≥ 55.0 OR ISM Prices ≥ 73.0 (i.e. at or above the prior print, no cooling versus the 70.0 forecast)

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

   PMI ≥ 55.0 or Prices ≥ 73.0: the case for a Fed hike is reinforced at its hardest point, input prices, yields gain room above 4.745%, and the dollar finally collects the reward it declined all last week. Gold and EURUSD both come under pressure; this is the only one of the three paths that breaks the bullish EURUSD bias.

2. In line

   Numbered threshold

   PMI between 53.0 and 55.0 AND Prices between 68.0 and 72.0

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

   PMI 53.0–55.0 and Prices 68.0–72.0: the data confirms manufacturing expansion while price pressure eases from 73.0 toward 70.0, not new enough to bend the yield path. Everything then returns to the standing theme: term premium weighs on gold, while EURUSD remains the cleanest vehicle for dollar decorrelation and holds its 1.1527 base.

3. Soft

   Numbered threshold

   PMI ≤ 52.5 AND ISM Prices ≤ 68.0 (slower growth alongside input prices cooling faster than forecast)

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

   PMI ≤ 52.5 and Prices ≤ 68.0: slower growth alongside faster price cooling puts a question mark over the very hike now being anticipated. A step back in real yields is the only condition sufficient to flip the bearish gold bias, and a softer dollar would carry EURUSD away from its old base.

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)BEARISHConviction 5/10

  Invalidated if: Gold clears and holds above the 4,063.1 resistance within a few hours of the release (more than half of the 30.6 H4 ATR14 from the 4,045.3 reference).

• EUR/USDBULLISHConviction 5/10

  Invalidated if: EURUSD loses 1.1508 after the release — the 1.1527 base broken rather than merely tested.

The most telling detail is how the dollar behaves if Prices lands right at 70.0: if DXY still cannot lift on a print that leans its way on rates, that is further evidence its problem is fiscal rather than cyclical.