US ISM services PMI — scenarios before, scorecard after

Before — what we said

The ISM Services PMI gauges US services activity; consensus expects 55.1, a slight dip from 55.4. We issue a call on Bitcoin only: its release-day range runs 1.21 times a normal day, while gold (0.92x) and EURUSD (0.91x) move no more than usual. Gold and EURUSD are therefore left out of the calls — their arrows in the scenarios are theoretical direction, not calls. The BTC call stays neutral in line with the running view, at low confidence because last time's neutral call was wrong.

The macro picture right now

Markets are pricing a hawkish Fed and fiscal risk: 10y real yield 2.88% (z +2.93), 2y yield 4.78% (z +2.43), MOVE 107.29 (z +3.13), HY spread 3.24% (z +2.21). With front-end yields at an extreme, a soft print has more room to pull the 2y down than a hot one has to push it up — the reaction may skew to the soft side. Equities are calm: VIX 15.31. For BTC, DVOL at 35.49 (z -1.01) is compressed just below 85,085 resistance while speculative positioning sits at 12.579% of OI (z +1.69) and funding is negative at -0.135‱: compressed volatility means a break can travel further than the 666 H4 ATR. Gold, FX and yields are Friday closes; only BTC (85,096) is live.

How the last print went

We do not yet have a history series for ISM Services, so its trend and typical range are unknown. Clues from the same September reference month lean soft: Conference Board confidence printed 81.9 against 89.0 expected, and consumer sentiment fell to 48.1, the second-lowest on record. Where this could be wrong: those are household surveys, while ISM polls businesses — the two have diverged for long stretches. Our record: on Sep 3 we called BTC neutral and were wrong (it moved 3.068% in 4 hours); on Aug 5 our bearish gold call was wrong and the neutral EURUSD call was right.

What it reaches, and how

BTC: ISM print -> 2y yield and USD -> Nasdaq risk appetite -> BTC. The event's lane matches BTC's dominant lane (Nasdaq r=+0.39 ±0.11), so the path is direct, with secondary routes via the 10y yield (r=-0.29) and DXY (r=-0.33). These routes align only on a hot print (yields up, USD up, both negative for BTC); on a soft print, lower yields support BTC while growth worries weigh on Nasdaq, leaving direction undetermined. Gold: its dominant lane is USD (DXY r=-0.49), so the event must route ISM -> yields -> USD -> gold, and the 0.92 ratio shows that route has not added release-day volatility. EURUSD: the risk-appetite lane reads r=+0.19, below the noise floor, and no axis dominates — no measurable path yet.

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

   ISM Services PMI <= 53.7

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

   A soft print would match September's consumer-sentiment cluster and pull the 2y yield off its extreme; the USD softens. For BTC, support from lower yields and drag from growth concerns offset, so the path depends on how Nasdaq reacts.

2. Numbered threshold

   ISM Services PMI between 53.8 and 56.4

   If this branch plays out:▬gold▬BTC▬EUR55%likelihood

   A print near consensus changes nothing in the real-yield story; BTC stays compressed beneath 85,085, and the running views on gold and EURUSD keep resting on fiscal pressure rather than on this number.

3. Numbered threshold

   ISM Services PMI >= 56.5

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

   A hot print confirms the economy can absorb high yields, reinforcing hawkish Fed pricing; yields and the USD rise together, and this is the only branch where both BTC channels align, so the move could exceed one H4 ATR.

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.

• BitcoinSIDEWAYSConviction 3/10

  Invalidated if: An H4 candle closes below 83,863 or above 85,650 within 4 hours after 14:00 UTC on Oct 5

If the print is soft, the tell is whether BTC follows falling yields or follows Nasdaq — that shows whether markets read weak data as rate relief or as growth risk.

Before — what we said

ISM Services PMI lands 14:00 UTC on 3 September, consensus 54.2 against a 54.1 prior — the market expects a flat number, so almost all of the price-relevant information sits in how far it strays from that band. We are issuing one call only, on Bitcoin: it is the sole asset whose release-day range exceeds a normal day (3.28% vs 2.83%, a ratio of 1.16 over 19 releases). Gold is deliberately excluded at a ratio of 0.91 — ISM Services days are narrower than ordinary days — and because on 5 August we ran a bearish gold call into this same event and were wrong. EURUSD is excluded at 0.82 and because H4 ATR14 is just 0.0015: a typical ISM reaction fits inside the 1.1575–1.1584 vice, where each edge already carries 9 touches.

The macro picture right now

The rate structure is at an extreme: 10-year real yields 2.44% (z +2.09) and the 2-year at 4.34% (z +1.98). That changes what a surprise means — the hawkish content of a strong print is already carried in ~70% September hike odds, so the hot branch has to be large to add anything, while the soft branch hits the most stretched pricing on the board and gets paid the most. Against that, HY credit spreads at 2.63% (z -1.42) show credit has not hedged growth risk at all: a weak number lands on a market with no insurance for that outcome, even with VIX at 16.34 after a 9.52% daily jump. For BTC specifically, implied vol (DVOL) 38.05 at z -0.78 means options are priced for a quiet day, while speculative positioning at 8.773% of OI has shed 6.105 points in five sessions — the leverage is already flushed, so any move has to come from real flow rather than a liquidation cascade. DXY 99.789 (+0.88% w/w) is firm while broad USD sits at z -0.93: dollar strength is currently a narrow-G10 story.

How the last print went

We have no free series source for this indicator family, so plainly: we do not know how many consecutive prints have run in one direction, nor the usual dispersion, and we have not established a sibling indicator sharing a reference period. The sharpest on-lane clue from the past week is the August Chicago PMI collapsing to 47.1 against a 57.9 forecast — a miss large enough to tilt the guess below 54.2. Where that clue can fail: Chicago is a regional, manufacturing-heavy gauge that has repeatedly diverged from national services, and it is not a forecasting coefficient. On our own record for this event: 5 August scored 1/2 — the bearish gold call was wrong (gold +1.175%), the neutral EURUSD call was correct (0.069%). That lesson is being applied here by declining to call gold at all.

What it reaches, and how

The event runs on the Risk-appetite lane. For BTC that lane is genuinely measurable (nasdaq r=+0.37 ±0.11, above the noise floor) but it is not the dominant one — dominance sits with USD strength (dxy r=-0.46), and that reading is level with VIX inside the error band, so neither channel deserves a crown. The consequence: the ISM number reaches BTC by two routes — a strong print lifts 2-year yields and the dollar and presses BTC down, a weak print both eases the dollar (up) and dents risk appetite (down). On the soft branch those routes cancel, which is why that branch stays neutral for BTC rather than bullish. For gold the path is blocked twice: the dominant lane is DXY (r=-0.57), which ISM only touches indirectly, while the event-lane correlation is nasdaq +0.35 — the wrong sign versus textbook, an unstable sign to bet on. For EURUSD the dominant lane is safe-haven/fear (vix r=-0.45), and a US growth gauge only arrives there after detouring through yields and equities.

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 Services PMI ≥ 55.0 (more than 0.8 points above the 54.2 consensus)

   If this branch plays out:▼gold▼BTC▼EUR26%likelihood

   Above 55.0 the print confirms services have not cooled and adds to an already extreme rate structure; the pressure on BTC arrives through the dollar channel rather than Nasdaq, and the natural destination for this branch is the 76,737 area before anything further needs discussing.

2. Numbered threshold

   ISM Services PMI in the 53.5–54.9 band

   If this branch plays out:▬gold▬BTC▬EUR46%likelihood

   Inside 53.5–54.9 the number adds nothing to the September rate path; BTC stays inside the 76,737–78,199 frame and all pricing weight shifts to Friday's payrolls — the heaviest branch, since consensus and prior sit just 0.1 point apart.

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

   Numbered threshold

   ISM Services PMI ≤ 53.4, breaking below the 54.1 prior

   If this branch plays out:▲gold▬BTC▲EUR28%likelihood

   Below 53.4 confirms the signal Chicago PMI already sent and hits the tightest spot on the board: easing real yields benefit gold and the euro directly, while BTC gets pulled both ways — HY credit spreads are the arbiter of whether the market reads it as easing or as weakness.

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.

• BitcoinSIDEWAYSConviction 6/10

  Invalidated if: Wrong if within a few hours of the print BTC closes an H4 candle above 78,199 (5 touches) or below 76,737 (4 touches).

The thing worth watching is not the headline but whether the 4.34% 2-year moves at all: if ISM prints strong and it stays put, September is fully priced and this event has spent its ammunition before payrolls.

Before — what we said

ISM Services lands at 14:00 UTC on 05 August, forecast 54.5 against a 54.0 prior — a modest step up, not a turning point. This is a mid-tier release and Friday's payrolls remains the real gate; the print's value sits in the prices-paid and employment subindices, the last read before payrolls. We keep the bearish tilt on gold in line with the running view: firm real yields under a hawkish Warsh continue to cap it while the haven bid keeps failing to pay. EURUSD stays neutral because the euro is the funding leg of the yen intervention, offsetting broad dollar weakness. We deliberately publish no call on Bitcoin: the transmission channel from a US services survey to BTC is effectively absent here, with BTC up only 0.38% over 24 hours and refusing to join the equity risk-on bid.

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 Services PMI >= 55.5 (more than a point above the 54.5 consensus and above the 54.0 prior)

   If this branch plays out:▼gold▼BTC▼EUR23%likelihood

   ISM Services at 55.5 or above: the services sector validates Warsh's hawkish turn and real yields firm further. Gold stays capped below 4,063.1 and drifts toward 4,046.3–4,022.6; the dollar briefly reconnects with rates and EURUSD leans toward 1.1499.

2. In line

   Numbered threshold

   ISM Services PMI between 53.6 and 55.4 (surprise of less than one point versus the 54.5 consensus)

   If this branch plays out:▼gold▬BTC▬EUR54%likelihood

   ISM Services between 53.6 and 55.4: the print changes no expectation and the market holds its wait-for-payrolls posture. Gold remains inside the firm real-yield regime and keeps a mild downward tilt around 4,059.9; EURUSD stays two-way within 1.1499–1.1544.

3. Soft

   Price moved most like this branch 2 of 3 assets agreed

   Numbered threshold

   ISM Services PMI <= 53.5 (below both the 54.5 consensus and the 54.0 prior)

   If this branch plays out:▲gold▬BTC▲EUR23%likelihood

   ISM Services at 53.5 or below, breaking the 54.0 prior: this is the branch that flips gold's short-term tilt, since it lets the market unwind some of the hawkish repricing and pull real yields down right before payrolls; EURUSD would then lean up alongside broad dollar weakness.

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 6/10

  Invalidated if: Gold trades above 4,087.4 within four hours of the release (more than one H4 ATR of 29.7 from 4,059.9) and holds above 4,063.1 into the US close.

• EUR/USDSIDEWAYSConviction 5/10

  Invalidated if: EURUSD closes an H4 candle outside the 1.1499–1.1544 band within a few hours of the release; a break either way means the two-way setup has been resolved.

The thing to watch is not the headline but the prices-paid and employment subindices — in a market where the dollar has been severed from rates (DXY 99.966, -1.52% weekly, against US10Y +4.48% monthly), the reaction is more likely to travel through gold than through DXY.