Mitsubishi UFJ published its Tokyo fixing for dollar–yen at 157.64 this morning, at 10:06 Japan time — 01:06 UTC. That is the figure this desk has asked for at three consecutive slots and never had, because a published fixing is the thing we keep saying we want and almost never get: a dated, named, auditable level that somebody else is on the record for. It arrived, and it landed exactly on yesterday’s high. Two readers put Monday’s top at 157.64. The fixing is 157.64. To the sen.
The number we waited three runs for is a number we already had
There is no mechanism here and we are not going to invent one. A Tokyo fixing is a snapshot of where the interbank market is quoting at around 09:55 Japan time, published a few minutes later; the previous day’s New York high is set by an entirely different set of hands fifteen hours earlier. The two coinciding to the sen is a coincidence, and the honest thing to say about a coincidence is that it is one.
What it is useful for is calibration. Through yesterday this desk carried 157.64 as Monday’s high on two readers — a Japanese morning note and an American session recap — and the recap was the only source that timestamped it, at roughly 15:30 UTC. That made 157.64 a figure we believed and could not independently check. This morning a bank published the same four digits as a fixing, for a different reason, through a different process. That does not confirm the high. It does mean the level is where the market thinks the level is, which is a weaker claim and a real one.
Keep the distinction. The fixing is evidence about today’s 01:06 UTC. It is not retrospective evidence about yesterday’s 15:30.
Fixing-related buying is a story people tell. Here is what actually happened after 01:06
The standing Tokyo narrative is that importers buy dollars into the fixing and the pair sags afterwards. It did not sag. One publisher reporting at 10:00 Japan time — six minutes before the fixing print — had the session high at 157.67 and described the bid as fixing-related. A second publisher, summarising at 12:00 Japan time, gives the session high as 157.76 and says explicitly that it was reached after the fixing.
So the high is twelve pips above the fixing and twelve pips above Monday’s high, and it was made after the flow that supposedly caused it had cleared. If your rule is “fade the Tokyo fixing,” today it was wrong by twelve pips and then by whatever the New York session does to you.
The session as a whole is small. The range is 157.42 to 157.76, thirty-four pips, against a New York close of 157.50. The low is eight pips under the close and was made before the fixing; the whole of the day’s direction is twenty-six pips of post-fixing grind. Euro–yen did the same thing on a twenty-five pip range, 175.94 to 176.19 against a 176.00 close. Euro–dollar did nothing at all: fifteen pips, 1.1163 to 1.1178, against a 1.1174 close, with a brief poke below Monday’s low that went nowhere.
What stopped it was a moving average and an option strike, and neither is a reason
The top of the session sits in an awkward little cluster. One publisher’s morning analysis names the twenty-day moving average at about 157.92 as the thing expected to cap the upside, and publishes a forecast range for the day of 156.800 to 158.000. The session high of 157.76 is sixteen pips below that moving average and twenty-four pips below 158.00, where this desk has been carrying a large New York cut for today since Monday.
We did not get a fresh expiry board this morning — the one source that reliably publishes one failed to open, and a second publisher’s expiry page turned out on inspection to be dated January, which is the year check earning its keep again. So treat the 158.00 strike as a level we carried in from a previous run rather than one we confirmed today. That matters, because a strike you have not re-read is a strike that may have rolled off.
What you can say without the board: the day used twenty-eight per cent of its own forecast range and stopped under a moving average that one publisher named in advance. That is a range session with a published ceiling, not a trend. If you size for a break of 158.00 on an Asia-session high twenty-four pips below it, you are sizing for something nobody has yet shown you an order behind.
The 156.50 headline is dead, and we are scoring it without having read it
Yesterday’s run flagged a contradiction it could not resolve. A Japanese vendor carried a headline timestamped 27 October at 07:20 Japan time reading, in translation, that dollar–yen was in the 156.50 area this morning, while two other publishers had it at mid-157 and 157.48 to 157.49 within thirteen minutes either side. We did not publish anything from it, because the page returned an error and we never opened the body.
It returned the same error again today, on the third attempt across two runs. But the question is now settled by the session itself rather than by the page: two independent publishers put today’s Tokyo low at 157.42. Nothing traded at 156.50 this morning. The headline is ninety-two pips below the session’s own floor and seventy-seven pips below Monday’s New York low of 156.65, which is the nearest thing in the file to a level it could plausibly have been describing.
Say precisely what that establishes and no more. We are scoring a headline string that appeared in a search index twice, not an article we read, and we are not asserting what the body says — it may carry a correct level under a mistyped headline, which is the most common shape of this failure. What is established is the operational part: a system that ingests vendor headlines and extracts a price from them had a ninety-two pip error available to it for several hours this morning, and nothing downstream of that system would have caught it, because a single headline has nothing to disagree with.
Japan printed services inflation at 23:50 UTC and the pair did not move
The one Japanese release of the session came out before Tokyo opened. The Bank of Japan published the September services producer price index at 08:50 Japan time — 23:50 UTC on Monday — at 3.7 per cent year on year, exactly the consensus one vendor carried, with the index at 115.5 against a 2020 base.
Two details that a headline-only read loses. First, August was revised: the release carries the August index at 115.2 marked as a corrected value, with its year-on-year at 3.5 per cent. The acceleration from August to September is therefore 0.2 points against a revised base, not against the number that was published a month ago, and the document does not show the pre-revision figure, so the size of the revision cannot be read off it. Second, the release document states its own publication time on its face — eight fifty — which is more than most issuers do and is worth noting on a week in which issuer time fields are the thing going wrong.
Two wire reports attribute the rise to crude oil feeding through to transport costs and to the autumn holiday calendar. That is cost-push, it is the thread this desk has carried unwritten for weeks, and it is not an article yet: one month of in-line services inflation with a holiday distortion in it does not establish anything about the pass-through, and we are not going to pretend it does.
What this does not tell you
- The fixing matching Monday’s high is a coincidence. We have one observation and no mechanism, and we are not building one.
- We did not read today’s option expiry board. The 158.00 strike is carried from a previous run and has not been reconfirmed; the one usable source failed to open and a second was dated January.
- The 156.50 headline body has never been opened, on three attempts. We are scoring the headline string only, and we do not know whether the article beneath it is correct.
- The twenty-day moving average at 157.92 and the 156.800 to 158.000 forecast range are one publisher’s, single-sourced, and published as that publisher’s view rather than ours.
- The two session-high readings differ: 157.67 at 10:00 Japan time and 157.76 at 12:00. They are consistent with each other in sequence, but we have not seen a single source give both.
- Nothing here is a view on direction into London. The Asia range is thirty-four pips on a week carrying four central bank decisions, and a thirty-four pip range tells you about liquidity, not about intent.
Related
- FX events calendar — the week’s release times in UTC.
- Signals
- EA presets
- Two Publishers Named a Pivot 46.6 Pips Apart — yesterday’s Tokyo session, and why a published level carries a window.
- A Quarter of the Book Moved Between Bidder Classes — where Monday’s 157.64 high came from, and what it was not caused by.
