Tokyo-area consumer prices for September land at 23:30 UTC, about a quarter of an hour after this goes out, and the consensus has core accelerating from 1.8 per cent to 2.4 per cent year on year. That is a six-tenths jump in a measure the Bank of Japan watches, and it will be reported as one. Before you let a filter treat it that way: one reader puts five and a half tenths of those six down to last September’s childcare and water-charge waivers dropping out of the year-on-year base. Ninety-two per cent of the acceleration is subtraction. The genuine new inflation in the consensus is five hundredths of a point.
The decomposition, and why it is the whole story
Tokyo abolished charges for a tier of childcare and waived the basic water charge last year. Those cuts pushed the index down in September 2025. Twelve months later they leave the comparison, and the year-on-year rate rises by the amount they had suppressed it — without a single price going up.
OANDA Japan’s preview puts childcare at roughly 0.30 points of the lift and water charges at roughly 0.25, about 0.55 points together. The consensus acceleration is 0.60. Divide one by the other and you get 91.7 per cent. Half the acceleration is childcare alone.
So the trade-relevant question is not whether core prints 2.4. It is whether it prints meaningfully above 2.4, because 2.4 is roughly what the calendar delivers on its own. A 2.4 is a non-event dressed as a six-tenths acceleration. A 2.6 would be the real surprise and it would look like a two-tenths beat. If your system sizes off the headline change rather than the deviation from consensus, this is the release where it gets the sign of its own conviction wrong.
August core-core, excluding food and energy, was 2.0 per cent and the headline was 1.9. Those are the series with no waiver in them, and they are the ones to look at tonight.
What the night actually did, and it was not about Japan
Dollar-yen traded 157.23 to 158.25 in New York on Fisco’s reading and closed at 158.25 — a hundred and two pips of range, finishing at the top of it. London had been higher: this desk published a 158.44 high at the pre-US slot, nineteen pips above anything New York managed. The yen is weaker, the pair is at the top of the week, and Tokyo opens with the dollar bid.
Now the part that matters more. Euro-yen fell from 178.58 to 176.77 on the same night — a hundred and eighty-one pips. The yen lost a hundred and two pips to the dollar and gained a hundred and eighty-one on the euro. It was the second-best major of the session, not the worst. Sterling-dollar lost sixty-seven pips and euro-dollar ninety-six, below 1.1300 for the first time since May 2025 on FXStreet’s reading.
If you are running a yen book and you read “dollar-yen up a hundred pips” as yen weakness, you have mislabelled the night. Nothing bought yen and nothing sold it especially hard. Europe got sold.
The dollar index high is a euro story, and the franc proves it
The dollar index traded near 102.10, through its June peak around 101.80 and its highest since April 2025, with the US ten-year at a twenty-four-year high above 5.30 per cent and the thirty-year at 5.65. FXStreet attributes the move to crude and the euro rather than to the odds on the Fed’s next meeting, and names China halting fuel exports and a report of a third carrier group as what moved oil.
The reasoning offered is that dearer crude hurts the energy-importing economies, which is Europe and Japan. Here is the problem with it. Dollar-franc fell from 0.8360 to 0.8292 on the same night — sixty-eight pips of dollar weakness against a currency belonging to an economy that imports essentially all of its energy and pays a lower policy rate than Japan does. The explanation that covers the euro and the yen does not cover the franc, and we are not going to pretend it does.
What a dollar index at a seventeen-month high tells you about dollar-yen, on this night’s cross rates, is less than it appears to. Two of its heaviest components did the work. One liquid funding currency went the other way.
On the ten-year, one open item narrows. We have had two publishers giving two different “highest since” framings for this level — 2002 and June 2007. A twenty-four-year high from 2026 is 2002. That is two readers for 2002 against one for 2007, and we still take the level and leave the superlative.
The 158.25 strike: broken before the cut, then exactly the session high and the close
One publisher described sizeable expiries at 158.00 and 158.25 for yesterday’s 14:00 UTC New York cut and suggested the pair might settle between them. Spot was at 158.44 before the cut, and this desk published that call as already broken. It was.
And then New York made a high of 158.25 and closed at 158.25. The upper strike, to the sen, as both the session high and the finish, seven hours after the option it belonged to had expired.
We are going to name that and refuse to make anything of it. Published strike-level calls on this channel are now one for three: a 157.00 pin that failed thirty-six pips below and closed thirty-eight above; this bracket, broken in London before its own cut; and this close, which landed on the strike after the strike stopped existing. Three observations with one apparent hit, arriving late, is a coincidence with a good story attached. We have no notional for either strike — the page that publishes them hides the sizes in an image, and the one that gave us a size once has not since — so we cannot even say whether the expiry was large enough to matter.
Two prints in the next forty minutes, and what they do to a window
Tokyo core CPI at 23:30 UTC and the Japanese unemployment rate at 23:50 UTC. Twenty minutes apart, both inside the first hour of Tokyo cash trading, in a pair sitting at the top of a hundred-pip overnight range with the dollar index at a seventeen-month high.
A fifteen-minute news window either side of each gives you two blocks with a five-minute gap between them — which is to say, in practice, a single fifty-minute hole from 23:15 to 00:05 in the one session where yen crosses actually have depth. Either accept the hole or treat the two as one event. Arming them separately buys you nothing and costs you the gap.
We do not have a consensus figure for the unemployment rate and are not going to invent one. The CPI is the release with the position in it, and on the decomposition above, most of what it says has already been decided by a subsidy that expired a year ago.
What this does not tell you
The 0.30 and 0.25 point contributions are one publisher’s estimate, read once, and they carry that publisher’s method with them. We have not reconstructed them from the Statistics Bureau’s own weights and we are not claiming the 91.7 per cent to a decimal. The shape of the argument — that most of the consensus acceleration is a base effect rather than new inflation — survives a fairly large error on either figure. The precise fraction does not.
The session range is Fisco’s. The 158.44 London high is our own earlier record from a single reader. Fisco’s own New York summary and its own reaction note disagree about what jobless claims printed last night, which is a reason to hold its figures loosely rather than to discard them, and our Systems desk has written that up separately this morning.
We have no crude level to publish. For the tenth slot running, a session has turned on the oil price and we could not get a Brent or a WTI figure we were willing to stand behind from more than one reader. Given that this desk carries a published claim about oil driving yields, that absence is becoming the story rather than a gap in it.
And none of this is a call on the yen. October odds on a Bank of Japan move were below twenty per cent on two readers yesterday morning, and a Tokyo CPI print that is nine tenths arithmetic is not what changes that. If you want the directional view, we do not have one. We have a reason not to size tonight as though a six-tenths acceleration were six-tenths of news.
Related
- FX events calendar — the 23:30 and 23:50 releases, and the rest of the Tokyo session, in UTC.
- Signals — range-versus-trend character in the Asian session.
- EA presets — news-window settings for clustered Tokyo-open releases.
