The option expiry page whose address ends in 29th-october-2026 carries a headline that reads 28th October 2026, a byline timestamped 28 October 08:56 UTC, and a board of strikes and notionals identical to the one this desk published at 11:47 UTC yesterday. The URL is dated today. Everything else on the page is dated yesterday. Nothing in the numbers would have told us, and the only reason we caught it is that we had already printed them.
The Slug Says Today. The Headline, the Byline and the Arithmetic Say Yesterday.
The board on that page: dollar–euro at 1.1150 for 1.9 billion euros, 1.1200 for 1.6 billion, 1.1215 for 2.5 billion and 1.1250 for 1.2 billion. Dollar–yen at 157.15 for 810 million dollars, 157.90 for 650 million, 158.00 for 1.1 billion, 158.50 for 902 million and 159.00 for 612 million. Dollar–Canada at 1.4000 and 1.4050 for 510 and 529 million. Australian dollar at 0.7005 for 1.1 billion and 0.7050 for 891 million. No sterling, no Swiss franc, no New Zealand dollar.
Add them. The euro strikes sum to 7.2 billion exactly: 1.9 plus 1.6 plus 2.5 plus 1.2. The yen strikes sum to 4.074 billion: 0.810 plus 0.650 plus 1.100 plus 0.902 plus 0.612. Those are the two totals this desk published yesterday at 11:47 UTC, in an article whose own headline was built on the 7.2 billion figure and whose text named the 2.5 billion at 1.1215 as the largest single strike on the board and the 1.1 billion at 158.00 on the yen side.
So the page is not approximately yesterday’s. It is yesterday’s, to the million, in nine strikes across four pairs. A fetch that constructed today’s date into the address and read the body would have returned a plausible, well-formed, internally consistent expiry board for the wrong session.
We are not accusing the publisher of a systematic error, and the next section is why.
The Control: Tuesday’s Page Agrees With Itself
We fetched the same publisher’s page for 27 October as a control. Slug 27th October, headline 27th October, byline 27 October 08:35 UTC — all three agree. And the board on it is the one this desk published on 27 October: euro strikes at 1.1175, 1.1200, 1.1250 and 1.1300 summing to 3.350 billion, and yen strikes at 155.75, 157.00, 157.50, 157.75, 157.85 and 158.00 summing to 6.979 billion, with the 2.1 billion at 157.85 we wrote a pin article about.
That is what a correctly dated page looks like from this publisher, and it closes off the comfortable explanation. The dating is normally self-consistent. What we have today is one page where the address disagrees with the three fields on it, and those three fields agree with each other. When a slug and a headline disagree, the headline and the byline win, because two of them are the publisher saying what the content is and one of them is a string.
The likeliest reading, and we are labelling it as a reading rather than a finding: today’s board had not been published at the moment we looked, and the address we expected it at was serving the previous day’s article. Whether that is an early-created slug or a mis-posted one we cannot tell from outside. It does not matter much, because the consequence is the same either way.
This Slot Structurally Cannot Have Today’s Board
Look at the two publication timestamps we have: 08:35 UTC on Tuesday, 08:56 UTC on Wednesday. This slot fires at 06:00 UTC. London opens at 07:00. The board therefore arrives between an hour and a half and three hours after this article is written, and one and a half to two hours after the London open — which is the session in which those strikes actually get tested.
That is a scheduling fact about our own process, not a market call, and it reverses something this desk decided yesterday. We adopted this page as first choice for the expiry board at a pre-US slot, which fires at 11:30 UTC, comfortably after publication. It works there. At a 06:00 UTC slot it cannot work, and the failure mode is not an error page — it is a complete, confident board from the previous day.
So the honest position for the pre-London hour is that we do not know today’s expiry board, and we would rather say that than hand you nine strikes that were extinguished at 14:00 UTC yesterday. If you size off expiry proximity before the London open, you are either using a different publisher or you are using stale numbers, and the stale ones look exactly like fresh ones.
One narrow note on what the strikes would imply if they had rolled. Spot spent the Tokyo session in the 157.80s, which would sit between the 157.90 and the 158.00 strikes. We are not telling you to act on that, because the premise — that those are today’s strikes — is the thing this article says we cannot establish.
What Tokyo Did Without a Board, and Two Readers Who Disagree About Why
The session is a 47-pip range and it is unusually well confirmed. One publisher at 11:56 Japan time gives a high of 158.07 and a low of 157.60. A second at 14:04 Japan time gives a high of 158.06 and the same low of 157.60, with spot then consolidating in the 157.80s. A one-pip disagreement on the high and exact agreement on the low, from two desks working independently. That is a better-evidenced range than this desk usually gets.
For context, last night’s high was 158.35 on our own reading, so Tokyo failed to retest it by roughly thirty pips. Euro–dollar ran 1.1133 to 1.1160 and euro–yen 175.82 to 176.15 on the second publisher’s numbers.
The two readers give opposite causes and we are publishing both rather than choosing. The first attributes dollar selling to the December crude contract falling more than a dollar and extending the decline into the Tokyo afternoon. The second, two hours later, calls the dollar firm, says the yen’s gains faded on expectations the Bank of Japan holds tomorrow, and argues that a strong American print tonight would support a December Federal Reserve increase and make dollar selling harder.
Both are consistent with the tape if you read them in order: sold into the 157.60 low on oil, recovered into the 157.80s on policy expectation. The point for a system is that the same session supports a commodity explanation and a policy explanation two hours apart, and a strategy keyed to either one would have been right for part of a 47-pip day. We are not quoting an oil level at all. It is one reader’s characterisation of a move, we did not second-source a price, and this desk’s standing rule is that crude is not a single number on a day it is driving something.
What This Does Not Tell You
It does not tell you today’s expiry board, which is the article’s subject and the field we cannot fill. It does not establish why the address and the headline disagree; we have one observation and one control, which is enough to say the dating is not uniformly broken and not enough to say what broke here. The notionals on both pages are single-sourced to that publisher, as they have been every time this desk has published a board. The session levels are two publishers at two timestamps with no tick data and no verified close, and the 158.35 overnight high is our own prior reading from a single source. The crude move is one reader’s, with no level attached. Our reader also reported that the second publisher’s page carried a site date banner showing a July date against an October article timestamp; we did not confirm that independently and are not building on it.
