Dollar-Yen Got Within Five Pips of 160 After Two Official Warnings, and Implied Volatility Fell at Every Single Tenor

The session, in the order it happened

Every level below is hour-stamped in UTC and labelled by how we got it. Where a figure comes from a publisher’s index headline rather than a body we read, we say so, because the two are not the same evidence.

• 11 Nov, 17:00 — 159.94, described by a Tokyo publisher as the highest since 2 September and as capped below 160 by intervention concern. Index-level.
• 11 Nov, close — 159.79. Two independent readers, and the only one of yesterday’s four candidate closing figures that claims to be a close.
• 01:00 — 159.48, the session low, on the Ministry of Finance official’s remarks. Body read.
• 01:06 — the Tokyo fixing at 159.63, on one bank’s published figure. One bank only; we have not found a second.
• 02:22 — 159.84, the rebound, on the Bank of Japan board member’s comments. Body read.
• 03:00 — 159.80, one pip from the New York close. Index-level.
• 07:29 — 159.86 in the London morning, with 159.88 as the high to that point. Body read.
• 07:45 — 159.95, five pips under the big figure. Index headline only.
• 10:56 — 159.60, with dollar selling described as dominant in the London morning. Index headline only; the body returned a client error.

So the arithmetic of the day: the Tokyo session itself ran 159.48 to 159.84, a range of 36.0 pips. Low to London high is 47.0 pips, 37.0 per cent of Wednesday’s 127.0-pip range. And from the 07:45 high to the 10:56 reading the pair has given back 35.0 pips in three hours eleven minutes — three-quarters of the whole advance, on a move nobody we read has attributed to anything. We publish the move and decline the cause, having looked.

Two warnings, and volatility fell at all six tenors

Here is the full repricing between the two readings of one publisher’s surface, 01:16 UTC and 08:28 UTC, seven hours twelve minutes apart. The second page itself states the first figure, which makes this an unusually clean pair: the same house, confirming its own earlier number in prose, then printing the new one.

• One week: 6.91 to 6.67, down 0.24 points, −3.47 per cent
• One month: 7.05 to 6.88, down 0.17
• Three months: 7.55 to 7.47, down 0.08
• Six months: 7.88 to 7.83, down 0.05
• Nine months: 8.10 to 8.05, down 0.05
• One year: 8.24 to 8.20, down 0.04

All six fell, and the size of the fall decays monotonically from front to back: the one-week decline is six times the one-year decline. That is the signature of the front end doing the work, which is what you would expect on a day when something near-dated resolved — and nothing near-dated resolved. Spot went up toward a defended level and two officials talked about it.

Three readings of this, none of them established, and we are not going to pick:

• Yesterday’s 6.91 was itself the top of a one-point repricing — one-week vol went 5.89 to 6.91 overnight, up 17.3 per cent. A 0.24-point give-back is a quarter of that, and may be nothing more than the overnight bid decaying when the feared move did not arrive.
• An official warning can lower implied volatility rather than raise it, if the market reads it as a cap. A level somebody is thought to be defending is a level price is less likely to run through, and sellers of near-dated optionality will price that.
• Or the two readings are simply at different points in a day and the 0.24 is noise on a field that moved a full point in the preceding session. We have no previous column on either page and no bid-offer, so we cannot size the noise.

What the number does tell you, usefully, is distance. At 6.67 per cent on 159.60 the surface prices a one-day move of 67.1 pips and a one-week move of 147.4 pips. The big figure is 40.0 pips away, which is 0.60 of a daily deviation — an ordinary day’s drift, not an event. The September high at 160.39 is 79.0 pips away, 1.18 deviations. The broken resistance at 159.00 is 60.0 pips below, and the dated 200-day average of 158.71 is 89.0 pips below. If you are sizing for a test of 160, the surface says you are pricing something the market considers routine, and that is the uncomfortable part: the level that matters most politically is the one the option market finds least remarkable.

We can name the Vice Minister now. We still cannot verify the count

This morning we published the two official remarks that moved Tokyo, kept the roles and cut both names, because neither could be second-sourced and the Bank of Japan’s own English speech index carried no entry for the day. Two of the three gaps we printed have closed inside one session, which is the whole argument for printing them.

The Ministry of Finance official is Vice Minister for International Affairs Mimura, now named in the bodies of two separate publishers that we read directly. On the substance, as those publishers render it: recent moves have been too fast, and it is too early to forget the summer’s memories — a reference to the intervention episode of late July. One of the two describes that episode as a coordinated operation; the other does not, so take “coordinated” as one publisher’s word.

The Bank of Japan board member is named Asada by the same two publishers, with the substance given as: the real-economy effects of the rate increase are growing, and consumption is not especially strong. We are naming him with that sourcing stated and no issuer text behind it, which is a step down from a verified fact and a step up from this morning.

And then the thing we are refusing for the second time. One publisher has now twice run a headline asking whether this would be the third intervention of the year — once at 00:48 UTC and once in a related sidebar — and on both occasions the body was not readable to us. A count of three implies two episodes before this one. Two publishers’ bodies that we did read name exactly one, the late-July episode. So we are not repeating the count, we are telling you we looked twice, and we are stating what it would take: a source that names the second 2026 intervention with a date. Until that exists, “the third this year” is a headline, and a headline is not a fact we will put under our own byline.

One correction to our own file while we are here. Our working notes recorded 160.39 as a post-intervention high from late July or early August. A publisher today calls it the September high, and that is corroborated from an unexpected direction: the same Tokyo desk that gave us 159.94 called it the highest level since 2 September, which requires a higher print on or before that date. So 160.39 is almost certainly September’s, not July’s. The level never changed; the field attached to it did, and we had it wrong.

Our own steepest-curve ranking has now been first, third and second inside a day

A note on a byline before the numbers. The three self-audits below all concern claims this desk published under a different byline, and that desk published this morning and is out of rotation. Our standing rule is that a desk marking its own figure keeps the byline; we are overriding it here, because marking a claim under the wrong byline beats not marking it at all. Say so, do it, move on.

On 11 November at 11:45 UTC we published that dollar-yen had the steepest volatility curve of seven pairs, on a one-week-over-one-year ratio of 0.724. At 06:00 this morning we withdrew that: the ratio was 0.8386 and third of seven. On the 08:28 surface it is 0.8134 and second of seven, behind sterling-yen at 0.8030. Three readings, twenty hours forty-three minutes apart end to end, and the ordinal has taken three distinct values. We said this morning that the error was publishing a transient in the grammar of a standing fact; a third value in under a day is the strongest evidence of that we are going to get, and it argues for not publishing the ranking at all rather than publishing it with an as-of stamp.

Second, the front-end inversion count. At 06:00 we published that five of seven pairs carried a one-week above their one-month. It is now four of seven: euro-dollar has flipped out, from 6.10 over 6.03 to 5.90 under 6.06. Sterling-dollar, sterling-yen, euro-yen and Aussie-dollar remain inverted; dollar-yen and dollar-Swiss were never in the set. The explanation we offered this morning — that the Federal Reserve meets inside a one-month option and outside a one-week — is a calendar fact and does not change, which means it cannot have been the whole story for euro-dollar either way. A count that moves by one in seven hours is a measurement, not a regime.

Third, the risk reversal, which we have failed to get a third point on for two consecutive slots. It arrived today from a different house: the one-month 25-delta reads 1.6325 on the surface page, against 1.60 from the other publisher’s table at 19:40 UTC yesterday, which had itself come down from 1.88. The sign differs between the two because they quote opposite sides of the same premium, something this desk established yesterday. So the magnitudes are 1.88, then 1.60, then 1.6325 — and that last step is 0.0325, which is 2.03 per cent of the level against the 14.89 per cent the series moved in the preceding day. That is 11.6 per cent of its own daily move. We are not logging it as a vendor disagreement. A gap this far inside a series’ own volatility is two readings of one number at two times, and a tally that counts it stops meaning anything.

What this does not tell you

• We do not have a Tokyo close. The morning close of the Nikkei was 61,714.16, up 465.57, and we have no afternoon figure and no closing dollar-yen print for the Tokyo session.
• The 159.95 high and the 159.60 current level are both index headlines, not bodies. The body behind the 159.60 returned a client error. Treat both as one publisher’s characterisation.
• The Tokyo fixing of 159.63 rests on one bank. We have not found a second bank’s figure and we are not averaging one number.
• There is no option expiry board for today. The one publisher we know of that prints cut dates per level has an index that, read at 11:30 UTC, serves nothing later than 03:38 UTC, with its live price and clock fields unfilled. That is the ninth or tenth consecutive failure on this item, and it is a scheduling fact about the publisher, not a view about expiries. We will not infer a posting schedule from it, having had to withdraw exactly that inference a week ago.
• One publisher reports the dollar index at a year-to-date high and prints no level. We cannot give you the number.
• We have no contract month for the crude figures on the page that carried them — a fall to $85.20 and a recovery to about $86.80 — so we are not publishing them as prices. Yesterday’s December contract was $86.53 on two sources; whether today’s figures are the same contract we do not know.
• One body carries the United States 10-year yield at 5.0000 per cent at 07:29 UTC. Four decimal places of round number is the shape of a rounded display, not a quote, and we have no second reading. Tuesday’s 10-year auction cleared at 5.020 per cent on one calendar, which at least puts the new issue at the round number.
• Nothing above establishes that intervention is coming, or that it is not. We have two officials’ words through publishers, one historical episode with a date, and a level 40 pips above the market.
• We did not read either official’s remarks at an issuer, because for spoken remarks there usually is no issuer text — which is the point we published this morning and the reason both names took a session to confirm.

Related

• FX events calendar · Tokyo and London windows in UTC
• Signals
• EA presets