With 150 minutes to run until the New York cut, the biggest notional on today’s option board sits 40 pips above anything the market has traded. Dollar–yen made 158.603 in the Asian morning, fell to 158.028 after the twenty-year auction, and was churning at 158.05 in early London. The board carries $1.1bn at 159.00 — 48.06 per cent of the dollar–yen total, with nothing above it. Yesterday this desk pre-registered that 159.00 would not trade before the cut. The claim is still live, and three separate things have moved against the strike since the Tokyo open.
The distance is 40 pips, and it has been widening all session
Take the numbers in order. The session high is 158.603, from one publisher’s order-information page timed 17:00 JST. That leaves 39.7 pips to the strike. The session low is 158.028, from the same page, and the low is 97.2 pips below the strike. The whole day has happened inside 57.5 pips.
A second publisher, stamped the same minute, has the pair at 158.05 and calls that the day’s low. So the direction of travel through Tokyo and into London is away from 159.00, not toward it. If you are sized for a pin, you are sized for something that has spent nine hours moving in the wrong direction.
That is not a forecast. It is the arithmetic of where the strike is relative to where price has been, and it is the only part of an expiry trade you can actually measure in advance.
The publisher that said 159 was possible has moved its own range below it
At 07:34 JST this morning, one Tokyo desk headlined its note with the view that a further rise in crude could see 159 tested. At 08:14 JST the same publisher gave a forecast range for the day of 157.800 to 159.200 — a 140-pip band with the strike inside it.
At 16:21 JST, that publisher’s evening note gives a forecast range of 157.600 to 158.700. The top came down 50 pips, the bottom came down 20, the band narrowed by 21.4 per cent, and — this is the part that matters — the top of the range now sits 30 pips below the strike. The same house, on the same day, moved from a range containing 159.00 to a range excluding it.
We have asked for eight weeks whether a publisher’s technical level set survives its own morning. Today it did not, and the revision went in the direction the price went. Treat a published intraday range as a description of the last hour, not of the day.
Nothing in the visible order book sits within 45 pips of the strike
One Tokyo publisher’s order information, as of 17:00 JST, lists sell orders at 158.300, 158.450, 158.500 and 158.550, and buy orders at 157.600, 157.700, 157.800, 157.850, 157.900 and 158.000. The highest sell order on the page is 158.550 — 45 pips below the largest notional expiring today. There is no listed order at or near 159.00 at all.
Say plainly what that is and is not. It is one publisher’s retail book, not the interbank book, and the absence of a retail sell order at 159.00 is not evidence that no bank has one. This desk retired two other vendors’ dollar–yen order-book characterisations and that retirement has not lapsed; we are reading this one as a sentiment distribution rather than as supply.
Read that way it is still useful, and it points the same way as everything else: the orders people have actually placed are clustered 45 to 100 pips below the strike that carries half the board. The same page attributes the heavy topside to repatriation flow after the auction and to speculation that the Bank of Japan lifts its FY2027 price view next week.
The cut minute also carries a US release, and its consensus is 70 basis points wide
The expiry sits at the New York cut. Two readers put September pending home sales at 23:00 JST, which is 14:00 UTC — the same minute. The consensus for it is not one number: one reader has it at zero per cent against a prior of plus 0.3, and another has it at −0.7 per cent. Seventy basis points apart, on a release landing in the minute the options expire.
An hour before that, at 12:30 UTC, housing starts and building permits print together. Starts are 1.305 million at one reader and 1.30 million at another, against a prior of 1.275 million; permits are 1.39 million at one reader and unmentioned at the other. Those two agree. The 14:00 release does not, and it is the one that collides with the cut.
The practical consequence is about window width, not direction. If your news filter blocks a window around 14:00 UTC, it is now covering two unrelated events at once, and the one you did not schedule for has a consensus you cannot state to better than 70 basis points. Widen the window or stand aside; do not size as if 14:00 were an expiry minute with nothing else in it.
The board for tomorrow is not where our own correction said it would be
This morning this desk retracted a scheduling claim and replaced it with a pre-registered test: if the board titled for a given day is published at 07:51 UTC the day before, then a board titled 21 October should exist by 07:51 today, and a check at 11:30 would find it. That check has now run, 223 minutes after the hypothesised publication time, and it found nothing. The dated URL for a 21 October board did not return a page, and a search restricted to the publisher’s own domain returns the 13th, 16th and 20th October boards and no 21st.
Two soft negatives are not a confirmed absence — one failed fetch is a failed fetch, not a 404. But this time the test could have returned a positive, which is exactly what this morning’s version could not do, and a negative from a test with power is evidence. The honest conclusion is narrower and more useful than either of our previous claims: the one-day lead is real for the instance we read twice, and it is not a cadence you can schedule against.
A job keyed to “yesterday’s 07:51” gets the board on some days and silence on others, and silence from an expiry feed looks identical to a day with no notable expiries. That is the same failure shape as the undated URL we published this morning: it does not error, it just stops being right.
We said the auction moved 15 pips. A later reader says 57.5.
Five hours ago our own Macro piece reported that dollar–yen moved 15 pips on the record twenty-year auction result. Two Tokyo readers now attribute more than that to it. One has the pair softening into the 158.20s at 13:10 JST with fiscal-deterioration worries easing and the yen bought on the back of the auction; the other has the pair falling from 158.603 to 158.028 — 57.5 pips, 3.83 times our figure — with the decline linked to repatriation flow after the sale.
Both can be true, and the reason is the window. Our 15 pips measured the minutes around the 03:37 UTC result. Their 57.5 pips is a session decline attributed to the same cause over the following nine hours. Neither number is wrong; they answer different questions, and only one of them is a measurement. A session-long move attributed to a morning event is an interpretation, however confidently a desk prints it — including when the desk is this one.
If you want the auction’s effect on your position sizing, the 15-pip figure is the one with a defined window. If you want to know why the pair is 55 pips lower than it was at breakfast, nobody in this sweep has shown you that, and three plausible contributors — the auction, lower US yields, and the Bank of Japan reports — are all in play with no way to separate them.
What this does not tell you
It does not tell you whether 159.00 trades. The claim is live at 11:30 UTC with 150 minutes to run and 40 pips to cover, and the score belongs to the Tokyo slot, not to this one.
The cut time is our convention. The board itself gives no time and no timezone for the New York cut; we read it as 14:00 UTC, and everything above about the collision with the 14:00 release depends on that reading being right. The board also carries no “last updated” field.
Every dollar–yen level here is a secondary reading. We did not reach an exchange or a primary quote feed, and the two publishers stamped 16:21 JST disagree with each other: one has the pair in the upper 158s capped by its 200-day average, the other has it at 158.05 making the day’s low. That is a 50-pip disagreement at the same stamped minute, and we are not adjudicating it. The 200-day average is cited by one reader with no value attached, so we cannot tell you where it is.
The order levels are one publisher’s retail book at 17:00 JST and are now four hours stale. The pending home sales consensus is two readers and we do not know which is the survey the market is actually trading. And the 21 October board may simply have failed to fetch; if it appears before the cut, this section is the part of the article that was wrong.
