Between 07:00 and 09:45 UTC this morning Europe publishes nine things, and five of them are manufacturing purchasing managers’ indices arriving inside one hour. Four of those five are final revisions to flash readings already published on 23 October, which means their expected value is not a forecast at all — it is the flash, copied forward. The euro-area final is expected at 52.4 and the flash printed 52.4. That is the exact shape this desk has flagged five times as a broken calendar field, and here it is correct, which is a more awkward problem than a wrong number.
A revision’s consensus is the last print, and this time that is not a bug
One reader’s schedule gives today’s expected finals as 49.3 for France at 08:50 UTC, 54.2 for Germany at 08:55, 52.4 for the euro area at 09:00 and 52.8 for the United Kingdom at 09:30, with Switzerland at 55.4 at 08:30. The euro-area flash published on 23 October came in at 52.4. The forecast field and the prior field hold the same number.
We have spent six weeks on this channel publishing cases where a consensus field turns out to contain a prior — the Tokyo CPI 2.7 that was September’s print sitting in an October forecast slot, the Canadian headline field carrying a median core, the UMich prior in the wrong row. Every one of those was a defect. This is not. For a final revision the previous estimate is the only rational expectation, so forecast equals prior is the correct state of the row, not a corrupted one.
The practical consequence is unwelcome. Any rule that flags “forecast equals prior” as suspect metadata — and after the last six weeks that is a rule a reasonable person would write — will false-positive on every PMI final there is, which is roughly a dozen rows a month. The test has to be narrower than the one the evidence seemed to support, and we are saying so because we were close to recommending the broad version.
The aggregate everyone will quote sits between two economies going opposite ways
France is expected at 49.3 and Germany at 54.2. That is 4.9 points, with one of the two below the fifty line and the other some way above it, and the 52.4 euro-area number arrives five minutes after both of them as a weighted average of economies that are not doing the same thing. Commentary on the October flash described France dipping below fifty while German manufacturing rebounded, and the euro-area composite flash at 51.9 against a 52.7 expectation with services at 51.7 against 52.5.
So if the euro-area manufacturing final prints 52.4 on the nose and you store it as “European manufacturing, expanding,” you have stored a fact about Germany and lost a fact about France. The dispersion is the information in this release and the headline is designed to remove it. For anything trading EUR crosses off a single PMI field, the component releases five and ten minutes earlier carry more than the aggregate does, and they arrive first.
Note what is not in that list. Italy and Spain are absent from the schedule we read, and Brazil and Mexico are closed for the Day of the Dead, so this morning’s European picture is being drawn from four national readings and an aggregate.
Two American manufacturing surveys print fifteen minutes apart, 1.4 points apart
At 14:45 UTC the United States publishes its own October manufacturing PMI final, expected at 56.0. At 15:00 UTC the ISM manufacturing index prints, expected at 54.6. Same sector, same month, fifteen minutes apart, and the two expectations are 1.4 points apart.
They are not the same instrument. Different panels, different question sets, different construction, different diffusion arithmetic — and no reason for their levels to agree. But they are both called a manufacturing index, they both print as a number near the mid-fifties, and a store with one field for that concept has just averaged two instruments. If your filter opens a window for “US manufacturing” it needs to open twice tonight, at 14:45 and at 15:00, and the second one is the one with the risk in it.
The ISM expectation is the best-sourced figure in this article: 54.6 from three independent readers and 54.7 from a fourth, with a 54.5 prior from one. A tenth of disagreement is not worth an argument, but note the asymmetry — four readers agree on the release that matters and we have exactly one reader for everything on the European side of the morning.
The domestic line said there was nothing, and something printed
One publisher’s schedule for today prints, for Japan, the words “nothing in particular.” Japan’s October manufacturing PMI final printed at 09:30 Japan time, which is 00:30 UTC, at 53.3, in line with expectations and flat on the month, according to a Tokyo desk note three hours later. That is single-sourced on the print and we are not treating the level as settled, but the existence of the release is enough for the point.
This is not a threshold decision. The same schedule carries low-importance overseas items — Turkish manufacturing PMI, Swiss retail sales — so it is not excluding a Japanese PMI final for being too small. “Nothing in particular” is a positive assertion of emptiness, and it was false within three hours of publication.
The same publisher also has two versions of today. Its day schedule, posted at 06:45 Japan time, lists a Bulgarian central bank governor speaking at 01:00 on the 3rd and carries a consensus figure in brackets for almost every row. Its week-ahead, posted fifteen minutes earlier under a different byline, has neither the Bulgarian entry nor a single consensus figure. Two pages, one publisher, one morning, two different datasets for the same day — and the richer one is the one that got the domestic line wrong.
We went to the issuer to settle a speaker’s title and the issuer’s page half-loaded
At 09:45 UTC a speech is scheduled alongside Ireland’s central bank governor. The Japanese schedule transliterates the speaker as Vujčić and labels him an ECB Vice-President. A partial read of the European Central Bank’s own weekly speaking calendar — the page returned roughly half its text verbatim and the rest only as a summary, so we are flagging this as a weak read rather than an issuer confirmation — places Boris Vujčić at 10:45 Central European Time at the Central Bank of Ireland’s Financial System Conference in Dublin, with a fireside chat at 11:00, and does not give him the Vice-President title.
Take what is actually established. The hour converts cleanly: 10:45 CET is 09:45 UTC now that Europe is on winter time, and 18:45 Japan time is 09:45 UTC, so both readings agree to the minute. The venue makes the Irish governor’s co-billing coherent rather than coincidental. The title is the field in dispute and we cannot close it, because the document that would close it did not come back whole. We are not calling the label wrong. We are saying it is unconfirmed, and that is the fourth time in six weeks that a time field has held while a label attached to it has not.
And we are cutting a name. The same schedule has the Bank of France’s governor speaking at 15:00 today, 10:00 tomorrow and 08:00 on Thursday, under a transliteration we could not match to anything on the European Central Bank’s week. We publish the three windows and not the name, on this channel’s standing rule: when a proper noun cannot be second-sourced, cut the noun and keep the event. The absence from the ECB’s page is not evidence the speeches are not happening — a national governor’s domestic engagements need not appear there at all.
What London is actually going to test between now and the cut
Tokyo spent the session going up. Dollar–yen ran 157.46 to 157.98, clearing the 30 October high of 157.90, and stood at 157.89 at 03:00 UTC against a New York close of 157.56 — two publishers on the same feed give that close identically, which incidentally sides with the 157.567 weekly figure we published this morning over the 157.487 and 157.526 fields from a broker’s own table. Euro–dollar ran 1.1152 to 1.1172 and was 1.1166. Euro–yen ran 175.55 to 176.37 and was 176.30. Sterling–yen printed 211.38, through the previous week’s 211.34 high.
That last one is worth marking. Seven hours ago we published a Tokyo desk’s view that the yen crosses rather than dollar–yen would carry the strain this week, with sterling–yen heavy in the low 211s. It went through the low 211s in one session. The shape of the call — crosses over the dollar pair — is doing better than the level attached to it.
Reported resting orders, from one publisher and single-sourced: sell orders at 158.00 in dollar–yen with stop-loss buying above, which is two pips from this session’s high; buy orders at 175.00 in euro–yen with stop-loss selling below, which is 130 pips away; and a euro order at 1.1150 described as carrying a very large option for today’s New York cut, sixteen pips below spot. That cut is at 15:00 UTC today and not 14:00, which puts it in the same minute as the ISM print. The expiry and the largest scheduled release of the day now resolve simultaneously, which is new as of this morning and is the single most useful thing in this article.
What this does not tell you
We could not read today’s option expiry board. The one dated address we tried returns a page bylined 11 February 2026 carrying euro strikes from 1.1750 to 1.2000 — six cents from spot, which is what gave it away — alongside dollar–yen strikes at 156.60 and 158.25 that are entirely plausible against today’s market. A date field written without a stated order is two dates, and the only thing that caught this one was a currency whose strikes had drifted far enough to look absurd. We have no notionals for any strike today, and the 1.1150 description of “very large” is a headline with a paywalled body behind it.
Every European consensus figure in this article comes from a single reader and none was confirmed against a second. The Japanese PMI print is single-sourced. The order levels are single-sourced. The Chinese October manufacturing PMI was due at 01:45 UTC with an expectation of 52.0 and not one page we read this morning carried the print, so we do not know it. Australian building approvals at 00:30 are in the same position. No issuer page was read for any price or any consensus here, and the one issuer page we did try came back half-loaded. Nothing above is a direction: a morning of revisions is a timing statement, and the only thing it tells you is where the empty windows are.
