The Bank of Japan’s governor is scheduled to address the National Securities Convention in Tokyo this afternoon, and the market is treating it as the last read on an October move before the 29–30 October meeting. The publisher that told us about it gives the time three ways in a single paragraph, and two of the three disagree by an hour. The Bank’s own release schedule for this week does not carry the speech. Neither does its list of monetary-policy speeches, whose most recent entry is 10 September. The one event on this morning’s Tokyo calendar with a documented number attached is a ten-year bond auction of about 2,600 billion yen, and that number comes from the issuing ministry.
The speech is at 05:35 UTC, or 06:35, and the same paragraph says both
One publisher, writing at 20:16 UTC yesterday, puts the address at “2:35pm Tokyo time / 0635 GMT / 0235 US Eastern time”. Convert the first of those yourself: 14:35 in Tokyo is 05:35 UTC. The second and third agree with each other — 06:35 UTC is 02:35 on US Eastern daylight time — and both disagree with the Tokyo figure by exactly sixty minutes.
You cannot tell from the page which is the error. If the local time is right, two of three renderings are an hour late. If the GMT stamp is right, the Tokyo time is an hour early. This desk has seen a sixty-minute offset at this publisher twice before and logged it both times as an anecdote; this is the third instance, and it is the first where the three inconsistent values sit inside one sentence rather than across two pages.
The practical consequence is not subtle. A news window is typically ten or twenty minutes wide. A one-hour error does not degrade it, it points it at an empty stretch of tape and leaves you unhedged through the actual event. Take the local time, convert it yourself, and if you cannot resolve the discrepancy, block both candidates. Two twenty-minute windows cost you forty minutes of the Tokyo afternoon. Getting it wrong costs you the speech.
The central bank’s own two calendars do not have this event
We checked both. The Bank’s release schedule for 5–9 October lists statistical publications: current-account balance sources and an output-gap release yesterday, market-operations and monetary-base data on the 7th, the Consumption Activity Index around 14:00 Tokyo time on the 7th, and the October Regional Economic Report at 14:00 on the 8th. No governor or deputy-governor appearance appears anywhere in that week. The next speech on the schedule belongs to a board member on 15 October.
The Bank’s separate list of speeches related to monetary policy stops at 10 September, a board member’s address in Fukui. The governor’s most recent entry there is 3 June, at a Tokyo meeting; the most recent deputy-governor entry is 27 August, in Saitama. There is no 6 October row and no securities-convention row on either page.
Two honest caveats, because they matter. The Bank posts speech texts after delivery rather than before, so an absence today is not evidence the speech will not happen. And an address to a securities-industry convention may simply not be filed as monetary-policy-related, which would keep it off the second page by design. Neither caveat rescues the operational point: if your event list is built from the central bank’s own calendars, you do not have this event at all. The only reason this desk knows about it is a commercial publisher, and that publisher cannot state the time consistently.
One in four, and it is one estimate
The same publisher puts the market-implied probability of an October hike at “roughly one in four”, sourced to “one estimate” with no feed named. The policy rate is 1.25 per cent and the meeting is 29–30 October, which we confirmed directly against the Bank’s own schedule rather than taking from the article.
That figure deserves a flag rather than a headline. This desk has tried and failed to obtain an October Bank of Japan probability at five consecutive slots. This is the first number we have, it is single-sourced, the estimator is unnamed, and we have no way to check how it is constructed. We publish it because having one bracket is better than having none, and we label it as one unnamed estimate because that is what it is.
Take it at face value for a moment and the sizing question is clear enough. One in four is not a coin flip and it is not nothing. It is the region where a governor’s tone can move pricing a long way in either direction, because there is room to travel both up and down. The same publisher notes that a deputy governor’s remarks briefly took the pair to around 157.5 before it recovered toward 158 — the name it attaches to those remarks is its own and we have not second-sourced it, so we leave the name out and keep the event. Fifty pips on a deputy’s comments is the calibration you have for what a governor can do at a quarter priced.
The auction is the one thing this morning that is properly documented
The Ministry of Finance announced it on 29 September: a ten-year Japanese government bond, auction 6 October, issue date 7 October, maturity 20 September 2036, offering amount about 2,600 billion yen. That is the issuing agency’s own announcement page and it settles a figure this desk had been carrying from a secondary reading.
Worth noting what the ministry’s October calendar does and does not say. The calendar page itself carries no amounts at all — it states that issuance sizes are published roughly a week before each auction date, which is why the per-auction announcement is the page you want and the calendar is not. The same calendar puts a thirty-year auction and a six-month bill auction on 8 October, a five-year on the 14th, a twenty-year on the 20th and a two-year on the 29th, all without amounts until their own announcements appear.
The asymmetry is the point of this article. The event with a precise, primary-sourced, double-checkable number is the bond auction. The event the market is actually positioning for has no entry on the issuer’s calendar and three mutually inconsistent timestamps at the only publisher carrying it. Your schedule’s coverage is inverted relative to its importance, and that is a property of where schedules come from rather than bad luck.
Where the pair actually is, and what did not break
Tokyo opened with dollar-yen around 157.90 after a New York morning high around 158.30, with the pair back above 158 in New York. Those two figures come from the publisher’s own headline text on its currency news index, because every individual article page we tried returned a 404 this morning; we are quoting the publisher’s approximations as approximations rather than pretending to a precision we could not read. A second publisher independently describes the pair recovering toward 158, which is the corroboration available.
That answers a question this desk left open yesterday. We published that the 158.40 first resistance and the 158.51 two-hundred-day line had both held through London, by 23 and 34 pips, with the 157.67 pivot breaking instead. Neither gave way in New York either. A high around 158.30 sits roughly 10 pips below the resistance and 21 below the moving average. The pair has now spent two sessions failing to reach a line it keeps being described as about to test, and the overnight high was made in New York rather than Tokyo for the second day running.
So the setup into the governor’s remarks is a pair pressed against overhead it has not cleared, in a session that will also absorb a 2,600-billion-yen auction, with a quarter of a hike priced and no reliable timestamp for the speech. The trade is not a direction. It is deciding whether you are in the market at all across two one-hour-apart candidate windows.
What this does not tell you
We do not know the governor will say anything about policy. Convention addresses are frequently ceremonial, and the market’s attention to this one is itself reported by a single publisher rather than observed by us.
We do not know which of the three timestamps is correct. We have reasoned that two agree and one does not; that is not the same as knowing which is right, and a reader who resolves it should trust their own conversion over ours.
The one-in-four probability is one unnamed estimate with no stated methodology. We did not find a second reader for it, and this desk has repeatedly shown that probability feeds for the same binary event can sit many points apart, so treat the number as a bracket rather than a level.
The dollar-yen figures in the last section are headline approximations from an index page, not levels read off a quote. We did not obtain an open, high, low and close for yesterday because the page carrying them was among those returning 404. We also did not publish a Japanese ten-year yield: the only reading we could reach was five days old, which is useless going into an auction, so we cut it.
And we have not checked whether other event calendars carry the speech. The claim here is about the central bank’s own two pages and one commercial publisher, which is what we read.
