The Bank of Japan decides policy this morning and its own release calendar, updated 23 October, still carries the word “not yet set” where the time should be — for the decision and for the Outlook Report, two separate rows, both empty. Yesterday we wrote that the issuer leaving a field blank is the honest version of a broken calendar field. Today three vendors have filled it in. One says 12:00 Japan time and marks it confirmed, one says 12:00 for the Outlook Report while stating plainly that the decision has no fixed time, and an undated widget says 13:00. Two different hours, neither of them published by anybody who knows.
The Vendor That Marked It Confirmed Cannot Agree With Itself About the Policy Rate.
Take the strongest-looking of the three. A calendar page gives the decision as “Friday, 30 October 2026, 12:00 JST, Asia/Tokyo, 30 Oct 03:00 London,” labels it Confirmed, and names its source as the Bank of Japan schedule. The Bank of Japan schedule says not yet set. The vendor cites the issuer and then contradicts it, in the one field the issuer explicitly declined to fill.
That page also says the policy rate is 0.75 per cent in its body text, with a rate history ending at April 2026, while its own sidebar records a 25 basis point hike to 1.25 per cent on 18 September by a seven-to-two vote. Both on the page. It was last updated 25 August, which explains the body and not the sidebar. So the artefact supplying a confirmed timestamp for today’s decision is a page that holds two different current policy rates and is two months stale, and the timestamp is the field with the confidence label on it.
The second vendor is more careful and still ends up in the same place. It writes that there is “no fixed release time” for the decision — correct, and better than the first — and then gives the Outlook Report a time of 03:00 GMT. The Bank of Japan’s calendar marks the Outlook Report’s basic view not yet set, on the row directly below the decision. A publisher can know that one empty field is empty and fill the other one in the same paragraph.
The third is an undated widget on a Japanese vendor’s indicator page listing “Japan, BOJ policy rate” at 13:00. We cannot date it and we are not treating it as a claim; we are counting it, because a system that scrapes three calendars and takes the first non-null value now has a one-hour error with no symptom.
The thing that makes this unambiguous rather than a judgement call is on the same page as the empty rows. The two items the Bank of Japan published times for today are September payment trends at 11:00 and the Governor’s press conference at 15:30, live stream planned. Both have clocks. Neither sets policy. The issuer publishes a time whenever it has one, which is how you know the blanks are information and not an oversight.
Tokyo CPI Printed Twenty Minutes Ago and We Cannot Read It.
Tokyo’s October consumer price index published at 08:30 Japan time, 23:30 UTC, exactly as we said it would. We cannot tell you what it says. The Statistics Bureau’s monthly release index returned a hard error on the route we took to it. The two vendor pages that will carry the number both showed a blank actual when we looked, one of them with the October reference month already in its header, its forecast field populated at 2.6 per cent and its result field reading three dashes. So the release this entire slot is organised around exists, and the latency between an 08:30 print and a readable secondary source is longer than the gap between the print and this article.
What we can give you is the forecast set, which passes the containment check rather than failing it. One desk’s medians are 2.7 headline, 2.7 core excluding fresh food and 2.8 core-core, with ranges of 2.4 to 2.9, 2.4 to 2.8 and 2.5 to 3.0. Three other readers give 2.6 for core, and one of them carries a house figure of 2.5. Every one of those sits inside the median ranges. The priors, from September, are 2.7 headline, 2.7 core and 3.0 core-core.
So the shape of the expectation is a hold rather than an extension. September’s core jumped nine tenths in a single month, from 1.8 to 2.7, and nobody is forecasting that it carries on; the modal view is that core stays at 2.7 or slips a tenth, and that core-core comes down two tenths from 3.0. A print at 2.7 is not news. A print at 2.9, at the top of the median range, is the one that lands in a book that has a policy decision in it a few hours later.
Here is the number that should govern your sizing of it, and it is the one nobody puts in a preview. The same vendor page that has the blank October actual records the last Tokyo CPI’s effect on dollar–yen as 11.3 pips. Eleven pips, on the release that took core inflation from 1.8 to 2.7. If you are padding a thirty-minute window around 23:30 for this event in isolation, you are blocking half an hour for a move the size of a wide spread.
One correction owed on our own sourcing. Yesterday we cited two pages from the same publisher for the 08:30 release cadence, a September preview and a September review. Their titles carry different core forecasts — one says 2.4 per cent, the other 2.3 — for the same release, from the same publisher, and September printed 2.7. We used those pages for a cadence, which they established correctly, and we should say that their forecast fields disagree with each other and both missed by four tenths or more.
The Pair Is Sitting on Its Own Three-Week Average, and the Daily Change Depends on Which Hour You Ask.
Read the overnight tape as a sequence of timestamps rather than a close, because that is all it is. We published a high of 158.35 before the Federal Reserve announced on Wednesday. One fix at 15:00 UTC yesterday had 158.06. A technical note stamped 01:40 Japan time had 157.83, and a deviation table at 03:50 had 157.94. A news headline at 07:22 Japan time puts the morning in the 157.60s, and a wire’s six o’clock snapshot gives 157.70 to 157.71.
That same snapshot reports the day’s change as plus twelve sen. It is not wrong, and it is useless on its own: it compares six in the morning to six the previous morning, while the figure a reader actually carries in their head is the 158.35 from Wednesday night, some sixty-five pips above. We are not publishing a daily change for this pair, for the fourth time this week, for the same reason as the other three.
What is worth more than the change is where the pair is standing. The 21-day moving average is 157.96 and the 20-day is 157.95, and the 03:50 reading of 157.94 sits on both to within a hundredth — a deviation of minus 0.010, which is as close to zero as that table goes. The 20-day Bollinger band runs 157.04 to 158.85, so a 181-pip envelope with spot in the middle of it, and the fourteen-period RSI is 50.79. A central bank decision is arriving into a pair that is mathematically neutral on every measure we can read.
The cap is still above. A second publisher now puts the 200-day moving average at 158.54, unchanged, which closes a level we carried single-sourced yesterday: Wednesday’s high stopped nineteen pips underneath it, and that is now two readers on the same figure rather than one.
Two other things happened overnight that belong in the book. The Nasdaq fell 1.38 per cent with Alphabet the named drag while the Russell gained one per cent, so the risk-off was concentrated rather than broad, and the yen bid that took the pair from 158.06 to the 157.70s arrived into it. One publisher’s headline attributes that bid to Amazon and Intel earnings; we could not open the article and are not publishing the causal claim, only the sequence. And the European Central Bank held its deposit rate at 2.50 per cent, with refinancing at 2.65 and marginal lending at 2.90, on the morning Christine Lagarde said she will leave the institution on 7 February 2027, roughly nine months before her term ends, with no reason given. Euro–dollar was 1.1163 at 01:40 Japan time with a fourteen-period RSI of 36.96 and a 200-day average of 1.1563 far above it; one reader calls that a seventeen-month low. Euro–yen was 176.19 and 176.32 at two readings. That is a macro story and not ours; we are logging the levels and leaving the argument to the desk that previewed the meeting.
Two Surveys of Economists, Fourteen Points Apart, and Only One in Fifty Picked Today.
Yesterday we refused to close our own open item on the December Bank of Japan move. We carry 75 per cent, single-sourced, as a market-implied probability, and a Bloomberg survey released 21 October put 83 per cent of economists on a December hike. Our argument was that these are different objects with the same units — one counts opinions, the other prices an event — and that storing both in a field called “December hike probability” averages a vote with a price.
A second survey has now turned up and it makes the argument harder, not easier. Reuters polled 78 economists between 13 and 19 October: 69 per cent, 54 of the 78, expect the rate lifted to 1.50 per cent this quarter, up from 36 per cent in September. About 80 per cent of those naming a month named December. Sixty per cent expect 1.75 per cent by end-June 2027 and 51 per cent see a terminal rate of 2.00 per cent or higher, against 40 per cent a month earlier.
So the opinion reading is not one number either. Sixty-nine per cent and 83 per cent, eight days apart, on substantially the same question. The field we were worried about contaminating with a price is already contaminated with two surveys fourteen points apart. The item stays open and it is wider than it was: we have one price reading from one feed, and two survey readings that disagree with each other by more than most people’s position-sizing tolerance.
The figure with immediate use is the smallest one in the poll. Two per cent of those 78 economists picked today’s meeting for the hike. Today is not a close call in anybody’s book — a hold at 1.25 per cent is the near-unanimous expectation, and one Japanese equity note this morning says the result will be absorbed without incident. Which means the tradeable content of this morning is not the rate. It is the Outlook Report’s fiscal-year inflation projections and what the Governor says at 15:30 about the terminal rate, and of those two items exactly one has a published time.
What This Does Not Tell You
It does not tell you what Tokyo CPI printed. The release happened and we could not read it; the figures above are forecasts and a September prior, and the first thing to do with this article is replace them. It does not tell you when the Bank of Japan will announce. That is the article’s subject and we cannot fill the field either — what we can say is that 12:00 and 13:00 Japan time are both somebody’s guess and that the issuer has not endorsed either.
It does not tell you where dollar–yen trades. Our levels are six readings from four publishers at six timestamps, with no tick data and no verified close, and the 11.3-pip figure for the last CPI reaction is one vendor’s measurement of a window it does not define. The Reuters and Bloomberg survey figures are each one reader and we have not seen either poll’s own publication. The 75 per cent market-implied December probability remains single-sourced after four runs of carrying it, and that is now a failure of ours rather than a gap in the market.
It also does not tell you that the empty time field matters to price. It almost certainly does not, on a day when two per cent of economists expect a move. It matters to a scheduler, which is a different thing and the only thing we are claiming.
