Our calendar said the Bank of Japan’s Regional Economic Report would land at 05:00 UTC today, one hour before this slot. It did, to the minute — the document itself gives its release time as 14:00 on 8 October, Japan time. Two of the nine regions carry an upward arrow and seven are marked unchanged. What the Bank does not do anywhere in the report is call those two arrows an upgrade; its own footnote defines the symbol as an acceleration in the pace of improvement. And at 06:17 UTC, seventy-seven minutes after publication, the Bank’s own index page for this report series still showed July as the most recent edition.
The event fired exactly where we said it would, which is worth saying out loud
This desk logged the 05:00 UTC time last night and set the judgement as this slot’s job. The release time printed on the document is 14:00 Japan time on Thursday 8 October, which is 05:00 UTC. Gaitame’s Tokyo outlook, filed at 08:12 Japan time this morning, carried the same 14:00 slot in its day-ahead schedule.
That is a clean score on a calendar entry, and clean scores on calendar entries have been scarce in this file. Over the past three weeks we have published a vendor with the wrong central-bank meeting date, a vendor with a release on the wrong day, a vendor whose “previous” field was the same month’s flash, and a vendor whose previous yield was 46.6 basis points from any security that traded. A scheduled event arriving when the schedule says it will is the baseline, and the baseline is what we had failed to get.
Two regions moved and the Bank of Japan does not call it an upgrade
Tohoku goes from picking up to recovering moderately. Shikoku goes from picking up moderately to picking up. Those two carry the upper-right arrow in the comparison column. Hokkaido, Hokuriku, Kanto–Koshinetsu, Tokai, Kinki, Chugoku and Kyushu–Okinawa are marked unchanged — seven of nine.
The wording changes are small enough that you could read them as editing rather than judgement, and that is why the Bank’s own definition matters. The upper-right arrow is labelled as an acceleration in the pace of improvement. It is a statement about the second derivative, not about the level. FXStreet, filing at 05:17 UTC, headlined it as an unchanged assessment for seven of nine regions, which is accurate and which deliberately does not say what the other two did. A wire headline we saw in search results framed the same two regions as an upgrade of the economic view. Both descriptions come from the same table; only one of them is the Bank’s own vocabulary.
If you have a news filter keyed to the words “upgrade” or “raises assessment”, it fired on this report. If you have one keyed to “unchanged”, it also fired, on the same report, from a different publisher, within the same twenty minutes. Neither publisher is wrong. The report supports both sentences, and a filter that cannot hold both is going to resolve the ambiguity by whichever feed it happens to be reading.
The issuer’s own page did not have its own report for at least seventy-seven minutes
At 06:17 UTC we fetched the Bank of Japan’s Regional Economic Report index. It returned July 9, 2026 as the most recent edition, with 2026 entries for January, April, May and July and nothing for October. One minute later we fetched the identical page with a cache-busting query string appended. It returned the October 2026 report, dated 8 October, at the head of the list.
This desk has a standing rule about appending a query string to vendor index pages, written after a news index served the same eight items with the same relative timestamps seven hours apart. We wrote that rule for vendors. This is the first time it has mattered at a central bank’s own site, and it is the case where it matters most, because the issuer’s page is the one you go to in order to stop trusting the vendors.
Read the sequence as an operator rather than a reader. A process that polls the Bank’s index for a new publication, on the plain URL, would have reported no October report at 06:17 — seventy-seven minutes after the report existed, after three separate commercial publishers had already filed stories about it, and twenty-two minutes after one of them had a price in it. The absence of an entry on an official page is not evidence that the thing did not happen. We already knew that about vendor lists. We did not have an instance at an issuer.
Search returned July’s report third for an October query, with a headline that contradicted both correct readers
Searching for this morning’s report returned, in the first three results, an article headlined to the effect that the Bank had maintained its assessment for all nine regions. Nine, not seven. Against the Bank’s own table that is simply false for today.
It is not false, though. It is dated 9 July 2026, 05:10 UTC, and it is about the July report — which did leave all nine unchanged, consistent with the July column of today’s own comparison table. The headline is correct about July and was surfaced by a query that named October.
We are recording this with the publisher unnamed in the body on purpose, because the mistake available here was ours and not theirs. The first question we put to every page we fetch is what its publication date is. That question is the only thing standing between this article and a published claim that a named vendor miscounted nine regions when the answer was seven. The near-miss is the finding: a date trap whose numbers are internally consistent, whose publisher is reputable, and whose headline would have handed us a vendor-fault story we would have enjoyed writing.
Tokyo took out 157.85, and the report moved nothing
Our open question into this slot was whether Tokyo would take out either end of yesterday’s 158.51 to 157.85 range. It took out the low. Minkabu’s Tokyo morning wrap, filed 12:07 Japan time, puts the session low at about 157.58 before the fixing — 27.0 pips through yesterday’s London low — with the pair recovering to the early 158s. Euro–yen, sterling–yen and Aussie–yen all printed their own lows in the same move and all rebounded, which makes it a yen move rather than a dollar move.
The 158.51 high was not threatened. Gaitame’s forecast range for the day is 157.400 to 158.600, with support at the Ichimoku conversion line in the 157.40s and resistance at 158.50 to 158.60 where the two-hundred-day average and the cloud’s lower edge sit. The 157.58 low stopped 18.0 pips above the bottom of that range.
And the report itself did nothing. FXStreet had the pair at 158.18, up 0.06 per cent on the day, seventeen minutes after publication — 60.0 pips off the morning low and inside yesterday’s range. A quarterly regional survey from a central bank three weeks before its meeting is not a trade, and we would say that before the price as well as after it. The useful part is the opposite of a signal: you now have a dated instance of two regions moving in the Bank’s own second-derivative language and the currency not caring, which is the correct prior for the next one.
What this does not tell you
We have not read the report’s regional detail beyond the summary assessments. The comparison symbols did not survive our first extraction of the Japanese document at all, and the Bank’s English summary page is where the two arrows were actually established. Anything about what is driving Tohoku or Shikoku specifically is not in this article because we have not read it.
We also cannot tell you how many regions the Bank itself considers to have been revised up, because the report does not count them and does not use the word. Two arrows is a fact. “Two upgrades” is a translation, and this article has tried not to make it.
The Tokyo low rests on one publisher’s morning wrap. We did not get a second reader on 157.58 and the figure is stated as about 157.58 in the source. The 158.18 reading is a different publisher at a different minute, so the 60.0-pip recovery is arithmetic across two sources rather than one series — which is the same criticism we have made of other people’s session ranges.
Nothing here speaks to October versus December in the Bank’s own pricing. The three readings we logged last night for an October move — roughly 10, 12 and 17 per cent depending on the instrument and the feed — have not been re-read this morning, and a report that did not move the exchange rate by more than six hundredths of a per cent is unlikely to have moved them either. We are not asserting that. We did not look.
Related
- FX events calendar — the 29 to 30 October Bank of Japan meeting and what sits before it.
- Signals — why a release that moves nothing is still worth timestamping.
- EA presets — news filters that have to survive two publishers describing one table in opposite words.
