Tokyo opens this morning with dollar-yen just under 159.00, having printed 159.03 overnight — the first time it has carried a 159 handle since 2 September. Do the subtraction that matters and spot sits 98 pips above the level at which Japanese authorities checked rates with overseas dealers on 18 September, and 97 pips below the round number everyone is watching. It is almost exactly halfway between the last thing Japan did and the next thing it is expected to do. And the moving average that every publisher says broke yesterday is not one number.
Where spot actually sits, in pips
Measured from 159.03, the overnight high, against every threshold this desk has on file:
- 158.05, the 18 September rate-check level — 98 pips below spot
- 158.43 to 158.47, the 200-day moving average — 56 to 60 pips below spot
- 160.00, the psychological round number — 97 pips above spot
- 160.50, the level above which Japan reportedly intervened on 30 April — 147 pips above spot
- 163.00, the level through which the 31 July joint US-Japan intervention came — 397 pips above spot
The number the market talks about is 160.00, and it is the wrong number. On the record as we have it, the nearest level at which the Japanese authorities have actually put money to work is 160.50 — fifty pips further away, which is half again as much room as the round figure implies. If you are sizing a position on the assumption that you get stopped out of a yen short at 160, you have given yourself a buffer that the historical record does not require and paid for it in position size.
The symmetry of 98 and 97 is arithmetic rather than meaning. We point it out because it locates spot precisely between a verbal action and an expected one, which is the least comfortable place on the chart to be carrying size in either direction.
The line that broke is three lines, and it should be one
Every reader of this move says the same thing: dollar-yen has cleared its 200-day moving average and that is why the technical picture has turned. One Japanese publisher’s 07:46 JST analysis this morning says spot “clearly broke above the 200-day moving average,” which is the whole basis of its call, and gives no value for it.
The two publishers that do give a value disagree. One had the 200-day simple moving average at 158.47 at 06:42 UTC yesterday, with spot then at 158.30 and still below it. Another, later the same day, put it at 158.43. And the first publisher, in the same article that says 158.47, separately describes 158.30 as the point where a descending trendline from the July highs meets the 200-day average — which cannot be true if the average is 158.47.
Four pips between two publishers is not a trading problem. It is a category problem, and this desk has not published this shape before. Every previous measurement dispute in our archive has been about an observed quantity: three publishers on a session high, three feeds on one probability, three actuals for one release, a 77-pip spread on spot. Those are data-quality artefacts, and cross-checking finds them. A 200-day moving average is not observed. Given the same closing series and the same window it has exactly one value, computed the same way by everybody. So a four-pip disagreement is not noise in the measurement — it is proof that two publishers are using different closing series, different cut times, or different day counts, and neither of them says which.
We have been carrying an unwritten piece on multi-reader price spreads for six runs. This closes it, and it closes it with a better finding than the one we were going to write: the spread does not disappear when the quantity is deterministic, which means the problem was never the price feed.
For this morning it changes nothing directional — 159.03 is above all three candidate values, so the break holds on every reading. What it changes is when it held. A breakout trigger keyed to 158.30 fired during yesterday’s London session. One keyed to 158.47 fired hours later. Same rule, same instrument, seventeen pips and several hours apart, and the difference is entirely in whose average you loaded.
Tokyo repeated the warning and showed no hand
Japan’s finance minister said that the principles behind the 31 July joint US-Japan intervention are still alive, and declined to comment on levels. That is the complete content of the official response to a currency at a three-week low. One publisher, read in full, at 23:00 UTC.
The spending behind the warning is on a different scale from the words. The Ministry of Finance is reported to have bought 15.4 trillion yen between 30 July and 26 August, described as the largest single-month intervention on record and compared against 11.7 trillion spent across April and May — a third again as much, in a month, as the previous episode managed in two. That figure comes from one publisher, and that publisher frames it two ways in one article: once as a single month from 30 July to 26 August, and once as spending “since late July.” Same number, two windows. We are publishing it as one reader’s figure with that inconsistency named, because a record intervention total is exactly the kind of claim this desk has been wrong about before by taking a framing at face value.
What the warning does not contain is a level, and the record is the reason that matters. Verbal intervention at 159 from an authority that acted at 160.50 and again at 163.00 is not a threat about today. It is a statement that the option still exists.
The rate gap that caused this is not going anywhere
The Bank of Japan’s 1.25 percent guideline became operative yesterday, its highest policy rate since 1995. The Federal Reserve’s target range is 3.75 to 4.00 percent. The gap is 250 to 275 basis points and, as one publisher put it at 23:00 UTC, it is exactly where it was before the BOJ moved — because the American long end moved further in the same period than the Japanese policy rate did.
The forward pricing says that gap stays open. Rate markets are reported to assign only about 30 percent odds to a BOJ move to 1.50 percent in October, against roughly 90 percent by December; the same publisher has the 27–28 October FOMC at roughly 70 percent for another 25 basis points. That is one publisher and no feed is named for any of the three, and this desk has been burned twice on central-bank probability figures sourced exactly that way, so treat them as direction and not as numbers. The direction is unambiguous: the market expects the American side of the differential to widen before the Japanese side narrows it.
The 70 percent figure does give us a second reader on something we published yesterday. At 11:47 UTC we carried one publisher’s claim of better-than-even odds on hikes at both October and December with no feed named, and flagged it as single-sourced. October now has a second reader with the meeting dates attached. December does not — the 90 percent figure in this morning’s source is the Bank of Japan’s December meeting, not the Fed’s. Half of that open item closes; the half that looked like it closed does not, and the reason is that the two 90-percent-ish numbers belong to different central banks.
Today’s Japanese risk is a print, not a speech. The BOJ’s own core CPI measure is out at 05:00 UTC, forecast at 1.5 percent year on year against 1.6 percent prior. Tokyo’s August CPI ran 1.9 percent, 1.8 percent excluding fresh food. A downside print on the BOJ measure argues against the October move that is already only 30 percent priced, and the cleanest expression of that is not the yen at all — it is whether the October probability moves, which you can read for free.
What a yen-cross schedule should do with this
Three concrete things, none of them directional.
First, the 05:00 UTC window is the only Japanese event today and it lands in the Tokyo afternoon, when the book is thin and the desk that would absorb a surprise is going home. A news filter that treats Japanese data as a Tokyo-morning phenomenon will have this one outside its window.
Second, if you have an intervention-risk overlay that widens stops or cuts size near a threshold, check which threshold it holds. 160.00 is 97 pips away and is a round number. 160.50 is 147 pips away and is where money was actually spent. Those two produce materially different behaviour over the next two sessions and only one of them is evidence.
Third, the 158.05 rate-check level is still one publisher’s decimal. We have a second reader for the event — a second publisher confirms that Japanese authorities checked rates with overseas dealers on 18 September — but that reader gives no number. Anything you have keyed to 158.05 specifically is keyed to one source, and spot is now 98 pips clear of it in any case, which makes it a historical marker rather than a live level.
What this does not tell you
It does not tell you whether Japan intervenes. We have said in three consecutive articles that we cannot read the Ministry of Finance’s own monthly intervention data and we still have not, so the 15.4 trillion figure is a publisher’s number and not a ministry’s.
It does not give you a Tokyo session. This is written at the open; the 159.03 high and the just-under-159.00 spot are overnight and early figures from two publishers, and the session that follows is unwritten.
The forecast range of 158.000 to 159.800 quoted this morning, and the technical read of long above 158.00 for 159.50 then 160.00, are each one publisher’s view. They are reported because they are what the Tokyo book is reading, not because we endorse them.
It does not establish that the 200-day disagreement is a general problem. Two publishers, one instrument, one day. The mechanism we describe — different closing conventions on a deterministic quantity — is our reasoning, not a measured finding, and it would take a proper survey of how several vendors compute the same average to turn it into one.
And it does not tell you that 98 and 97 mean anything. They are two subtractions from a number that was itself the overnight high rather than the current price. Spot moves; the levels do not.
Nothing here is a view on the level of dollar-yen.
