ECB — scenarios before, scorecard after

Before — what we said

The ECB publishes the Main Refinancing Rate at 12:15 UTC on Sept 10, with consensus at 2.65% from 2.40% — a 25bp step. The number itself is close to fully paid for; the actual price driver is the 12:45 UTC press conference, which shapes the next meeting rather than this one. We deliberately leave Bitcoin out of the calls: its release-day range runs 1.46 times a normal day, but that is a volatility measure, not a direction, and the only route from the ECB to BTC is the dollar detour (r=-0.45), weaker than gold's. On top of that BTC is pinned at 78,723 resistance, funding has been flushed and speculative positioning has dropped 15.9 percentage points of OI in five sessions — no positioning asymmetry is left to convert a dollar move into a BTC direction. Calls stay on gold and EUR/USD.

The macro picture right now

The market enters this meeting with two extremes pointing opposite ways. The 10y real yield at 2.42% (z +1.92) and the 2y at 4.34% (z +1.9) say the hard part of the global rate structure is already at cycle highs — a more hawkish ECB cannot create a fresh rate shock, it only confirms what is priced. Against that, speculative EUR positioning is still at -2.88% OI (z -1.05) despite 5.957 points of covering over five sessions: the book still leans short euro, so a hawkish surprise has squeeze fuel while a dovish one lands on people already short — the asymmetry sits on the upside. Broad USD at 118.748 (z -0.93) and DXY at 98.759, down 0.91% on the week, show the dollar channel is already open on the soft side. Risk is not what is running the tape: VIX 15.72, HY credit spreads 2.65% (z -1.29), MOVE 73.1 (z +0.02) — no event premium has been built into bonds ahead of the print, which means any reaction has to be constructed from scratch. Gold's own crowding is the tightest thing on the table: speculative positioning at 54.944% OI (z +1.27) will amplify any firm-dollar outcome.

How the last print went

The indicator series carries exactly one level change: 2.40% on Aug 29. The source only emits 12 observations, so earlier steps are invisible — do not read this as 'the rate rarely moves'. What it does say is that the ECB changed the level less than two weeks ago and the market already expects another change: this is a fast-running cycle, not a one-off adjustment. The closest same-lane clue is a rate-differential headline saying the ECB is expected to hike again but signal caution on subsequent moves — the classic 'hike then brake' template, and the reason the consensus branch carries the weight. Where that clue can fail: it is pre-meeting expectation, not the statement, and September is a staff-projection round — one upward revision to the inflation forecast neutralises the 'cautious' half. The same lane also carries the BOJ expected at 1.25% on Sept 17-18, so the hawkish impulse is global and the reward specific to the euro gets diluted. We have no scored prior round of this event, so there is no track record to calibrate against; confidence stays moderate.

What it reaches, and how

This event runs on the dollar-strength lane. For gold, the event's lane is exactly the lane currently in charge — dxy r=-0.56 ±0.09 — so the path is direct: a stronger euro pulls DXY down and cushions gold, a euro that fails takes away the last support gold has and hands pricing back to US real yields. Gold's release-day range is 2.25% versus 1.82% on a normal day (1.24x, n=12): real impact, moderate size, and the reason confidence on gold sits below the FX call. For EUR/USD the situation inverts in a way worth stating: none of our axes clears the noise floor for this pair, the dollar lane included — the transmission here is not statistical but definitional, since the ECB sets one leg of the exchange rate outright. The 0.74% versus 0.51% release-day range (1.44x) is the footprint of that mechanism, and it is nearly six times the current H4 ATR of 0.0015 (0.13%). The compressed 1.1619–1.1646 band is unlikely to survive the press conference; the question is not whether it breaks but whether the break holds.

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 1 of 3 assets agreed

   Numbered threshold

   Main Refinancing Rate at 2.65% or above AND the press conference leaves at least one more 25bp step open (or a direct move to 2.90%)

   If this branch plays out:▬gold▬BTC▲EUR22%likelihood

   The hike arrives with the door left open for another step, or with staff inflation forecasts revised up. This is the branch with positioning fuel: the book is still net short euro, so the reaction takes the shape of a squeeze rather than a repricing — fast, with most of the distance covered in the first hours. Gold gets pulled both ways here: a softer dollar supports it while the global hawkish impulse lifts real yields, and the two channels cancel.

2. Numbered threshold

   Main Refinancing Rate exactly 2.65% (+25bp), with data-dependent language and no commitment for the next meeting

   If this branch plays out:▼gold▬BTC▬EUR57%likelihood

   The template outcome: exactly 25bp, data-dependent language, nothing promised for the next meeting. The first move is a reaction to the news being out rather than to its content; the euro typically breaks the band and gets pulled back, because no new rate differential has been created. Gold receives nothing from the ECB in this branch and hands pricing back to US real yields, where the bearish bias is already running.

3. Numbered threshold

   Hold at 2.40%, or a hike to 2.65% paired with a press-conference confirmation that this is the terminal rate

   If this branch plays out:▼gold▼BTC▼EUR21%likelihood

   Either the ECB holds at 2.40%, or it goes to 2.65% and declares the terminal rate reached. This is the only branch that firms the dollar through the euro leg, and the only one that hits all three assets in the same direction, since all three hang on the same DXY hook. But because the book already leans short euro, the distance travelled will be shorter than an equivalent shock in the other direction.

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 6/10

  Invalidated if: EUR/USD closes an H4 candle outside the 1.1619–1.1646 band and the next H4 candle fails to pull price back inside — that is a genuine repricing, not noise around an already-paid-for level.

• Gold (XAU/USD)BEARISHConviction 5/10

  Invalidated if: Gold reclaims and holds above 4,373.4 (9 touches) within a few hours of the press conference.

The thing that matters is not the 12:15 UTC number but how the 12:45 UTC press conference describes October: one sentence leaving the next step open and one sentence closing it produce two entirely different euro markets, while 2.65% itself has already been paid for.