New York traded USD/JPY from 154.86 to 156.42 while Tokyo was asleep — one hundred and fifty-six pips, one publisher’s reading, and the first time 156 has printed in a sequence this desk has spent three sessions writing about at 155. Tokyo opens this morning with the figure still trading around 156, Japan’s national consumer price index under twenty-five hours out, and the Bank of Japan under twenty-eight. Sixty-six of sixty-eight economists expect a hike tomorrow. That is the problem, not the comfort.
The Number, and the Caveat That Goes With It
The overnight range we can give you is 154.86 to 156.42, and it comes from a single Japanese publisher’s morning note timed 07:06 JST. We could not second-source the high before writing: the pair page at our usual Western reader was still serving a pre-decision figure of 155.45 to 155.50, and the Tokyo desk notes that normally give us an independent high and low had not published for 17 September when we looked. So treat 156.42 as one reader’s number with a live risk of being a few sen out.
What is not in doubt is the shape, because three publishers describe the same path. USD/JPY was near 155.10 into the 18:00 UTC decision. It sold off into the 154 handle on the announcement. Fifteen minutes later it was around 155.44. By 18:35 UTC, with the press conference running, it was near 155.88, and it eased to roughly 155.66 ten minutes after that. Then it kept going. Whatever the exact high, the yen lost its Fed-day rally inside a quarter of an hour and spent the rest of the New York session giving up ground.
Three Sessions of 155. Then It Was Gone in One.
On Tuesday night this desk wrote that USD/JPY had taken 155.00 three times in thirty-six hours, that the first two attempts were handed straight back inside the session, and that the third one stuck at a 155.10 New York close. Yesterday morning we wrote that Tokyo had handed London forty-six pips. Yesterday’s Tokyo session traded a 154.93 to 155.49 range and the pair actually slipped below 155 during it, on what one Tokyo desk described as Bank of Japan hike expectations.
That entire structure was removed in about six hours. The level that took three attempts and thirty-six hours to clear is now more than a figure below the market.
The lesson is not that we were wrong about 155 — the sequence we described happened. It is about what a hard-won level is worth once the driver changes. Every one of those three attempts was driven by the US ten-year. The break that finally worked was driven by sixteen of eighteen Fed officials putting another hike on a dot plot. A level that took thirty-six hours to defeat under one driver took minutes under another, and if your system sizes resistance by how long it has held, it just paid for that assumption.
Japan Did Print Something, and Tokyo Did Not Trade It
We owe a reader a loose end. Two mornings ago we published a falsifiable prediction: that USD/JPY would not move more than about twenty pips in the ten minutes after Japan’s 23:50 UTC trade balance, because three sessions running every link in the yen’s chain had been American. At the next slot we could not mark it, because none of our sources publish at ten-minute resolution and we could not even read what Japan had printed six and a half hours later.
We can now read part of it. August exports rose 19.3% year on year against a forecast of 18.2% and a prior of 23.2% — a beat of just over a point, from one Tokyo desk note timed 16:36 JST yesterday. The trade balance headline and the machinery orders we still do not have.
So the prediction stays unmarked and we are not going to quietly let it lapse. What we can say is the weaker claim it rested on: Japan beat on exports, and the Tokyo session that followed traded a fifty-six pip range and finished lower against a dollar that then took the pair a hundred and fifty pips higher on an American event. Three sessions became four. We still cannot measure ten minutes, and we are increasingly of the view that publishing a threshold our own sources cannot resolve was the error, not the threshold itself.
Sixty-Six of Sixty-Eight Is the Problem
A Reuters poll of sixty-eight economists, conducted 1–8 September, had all but two expecting the Bank of Japan to raise the policy rate to 1.25% on 18 September. More than a third — twenty-four of sixty-six — expect a further move to 1.50% in October or December. Fifty-seven of sixty-four see at least 1.50% by the end of March.
Ninety-seven per cent is not a forecast, it is a fact already in the price. Which means tomorrow’s decision is not a trade. The decision cannot surprise you unless it does not happen, and a two-in-sixty-eight outcome is not something you size for — it is something you survive.
The trade, such as it is, sits in the conditioning language and in the press conference that follows. Note the asymmetry the poll creates. If the Bank hikes and says nothing about the path, the yen has been told it is getting 1.25% and it already knew. If the Bank hikes and points at 1.50% by year end, it is confirming what a third of the panel already expects — still not news. The genuinely unpriced outcomes are a hike with language that pushes the next one out, or no hike at all. Both of those are yen-negative into a pair that has just broken 156.
That is an uncomfortable shape. The event everyone is positioned for has its surprise risk stacked on the side that hurts the yen, at the top of a hundred-and-fifty-pip move that has already happened.
The Two Clocks That Matter Today
The Bank’s policy meeting begins today; the decision comes tomorrow, with no fixed minute — roughly 03:00 UTC on past form, and the Bank does not promise a time. Japan’s national consumer price index lands at 23:30 UTC tonight, under twenty-five hours from now and about three and a half hours before that decision. An RBA speech is scheduled for the same minute as the CPI.
Two things follow for a schedule. First, a national CPI print three and a half hours before your own central bank decides is not a data release, it is the last input, and thin Tokyo liquidity at 08:30 JST is where it lands. Second, the CPI and the RBA speech share a minute, so a filter keyed on JPY events alone will let an AUD event through the same window — check whether you hold correlated AUD/JPY exposure before tonight rather than after.
Between now and then: the Bank of England decides at 11:00 UTC, under twelve hours away, and US jobless claims and the Philadelphia Fed survey land at 12:30 UTC, under fourteen. On a morning when the yen has just moved a figure and a half, the London session is the quiet part of the day, and that is the part most likely to be mistaken for a trend.
What This Does Not Tell You
The 156.42 high is one publisher’s figure and we said so at the top rather than at the bottom. The 154.86 low comes from the same note.
We do not have an option expiry ladder for today. That is now the sixth consecutive slot at which we have looked for one and had nothing to report at the time of writing, and readers who size on strikes should assume they are trading without that map this morning.
We cannot tell you where the order book sits. The Tokyo order-book sources this desk used to quote were retired for USD/JPY after they proved unreliable, and we have not replaced them. A hundred-and-fifty-pip break with no readable resting-order information is exactly the situation in which a level invented by a chart is most dangerous.
The trade balance and machinery orders from 15 September are still unread, six days of trying. At some point that stops being a research gap and becomes an admission, and we are close to it.
And nothing here is a view on whether the yen goes up or down from 156. The Bank of Japan is going to raise rates tomorrow into a currency that has just weakened a figure and a half on somebody else’s central bank. That is a sizing problem and a timing problem. It is not a direction you can read off a poll.
