The Blank Cell Was Not a Fourth State. Two of the Three Blanks Are Under the Billion the Board Would Not Name — and the One Violation Misses by Six Per Cent.

Publicado: Atualizado: 2026/11/18 11:46 UTC

Two of the three dollar-yen options expiring at the New York cut carried no size band at all on the only structured board this desk trusts. A second publisher has now put a figure on every one of them, and the figures say the blank cells were ordinary. 160.50 is 946.31 million dollars. 159.75 is 570.76 million. Both sit under the one billion that is the board’s own lowest named threshold — which means a blank cell on that board is not, as this desk published at the pre-US slot of 17 November, a fourth state of unknown magnitude. It is a threshold test, and we can now say how wide its failure is: 60 million dollars on 1,060, or six per cent.

We called a blank cell a fourth state. It is a threshold test that fails by six per cent.

The claim being marked down is ours, published under this byline at the pre-US slot of 17 November: that the board defines three size labels, all of them floors on the large side, uses two more it never defines, and therefore leaves a blank cell as a fourth state sitting next to two undefined small ones. The practical reading we drew was that a blank tells you nothing. That was too strong, and the correction comes from putting a second publisher’s numbers against the first publisher’s labels on the same cut date.

Three dollar-yen strikes expire at 15:00 UTC. The notional board gives all three:

  • 160.50 — 946.31 million dollars. 125.0 pips above spot. The banded board carried this strike for this date and left its size blank.
  • 159.75 — 570.76 million dollars. 50.0 pips above spot. Also carried, also blank.
  • 158.50 — 724.36 million dollars. 75.0 pips below spot. Not on the strikes we were carrying for this date at all.

The two blanks are 0.946 and 0.571 of a billion. Against them stands the counterexample this desk found at the Tokyo slot of 17 November: a 160.00 strike that the banded board also left blank and that a notional source put at 1.06 billion dollars. So three observations of a blank cell now exist, two under the threshold and one over it by 60 million on 1,060 — which is 1.060 times the floor, not two times or five times it.

That is a different object from a fourth state. A blank cell is weak evidence of “under one billion”, and its one known violation misses by six per cent. A store that reads a blank as small will be right about two strikes in three and wrong by a rounding error on the third. A store that reads a blank as unknown, which is what we argued for, throws away information that is mostly correct. Neither reading is free, and the second is the one we recommended.

The band confirms from the other end, and this is the first time we have been able to check one.

The euro leg is the cleaner test, because it exercises the top of the vocabulary rather than the bottom. The banded board flagged a 1.1025 euro-dollar option for this cut with its heaviest label, which the page defines as three billion currency units or more. The notional board gives that same strike as 4.33 billion euros.

4.33 against a floor of 3.00 is 1.443 times the floor. The band is honest. This is the first occasion in our file on which a size band from one publisher and a notional figure from another have been put against each other on the same strike and the same cut date, and the band survived. It is one observation and we are not generalising from it, but it does change how we will read that board: the labels appear to mean what they say, and the problem with the board was never that its bands lie. The problem is that it leaves the interesting cells empty.

Worth saying plainly, because it cuts against us: the strike agreement is also what makes the cut date believable. The notional publisher’s standing weakness in this file is its labels and its metadata rather than its numbers, and a cut date is metadata. We are leaning on its numbers, which are the field it is reliable in, and taking the date from the fact that two of its three dollar-yen strikes and its euro strike are exactly the strikes a different publisher independently dated to this cut.

Size runs the wrong way against distance — and that was a real property, not a gap in the data.

Under this byline at the pre-US slot of 17 November we wrote that the size information on the banded board was inversely related to its relevance: the only strike the publisher would characterise sat nearly two and a half deviations from spot while the nearest one carried nothing. The obvious objection is that this was an artefact of missing data — that the near strikes were sized too and we simply could not see it.

They were sized, we can now see it, and the inversion is still there. On one-week implied volatility of 5.77 per cent a standard day is 57.88 pips, and on that scale:

  • 159.75, the nearest strike at 0.864 of a standard day, is the smallest at 570.76 million.
  • 160.50, the farthest at 2.160 standard days, is the largest at 946.31 million — 1.658 times the nearest.
  • 158.50, at 1.296 standard days below, sits in between at 724.36 million.

Total dollar-yen notional across the cut is 2.2414 billion dollars, and the single largest strike is 42.2 per cent of it. The ranking by size is the reverse of the ranking by proximity on all three. So the inversion we described was a property of the board and not of the holes in it, and the only way to establish that was for somebody to print the numbers. A table that sorts expiries by distance and reports the nearest would put 159.75 first and be reporting the smallest option on the cut.

What none of this supports is a pin argument. The nearest strike is 50.0 pips away with the cut at 15:00 UTC, and it is the smallest of the three. The heavy end of the board is 125.0 pips out at better than two standard days. If you are sizing exposure into that cut, the honest input is that there is no concentration near price to trade against, which is a reason to care less about the cut than the strike list makes it look.

Volatility came in, the range did not, and the 200-day finally has a value and still has no date.

One-week dollar-yen implied volatility is 5.77 per cent as of 16:34 Tokyo time, which is 07:34 UTC, down from 6.07 per cent in the Tokyo morning on the same publisher’s reading. That is 0.30 points, and it shrinks a standard day from 60.89 pips to 57.88 — 3.01 pips of scale removed before the New York session begins. The term structure is ordinary: 5.77 at one week, 6.75 at one month, 7.37 at three, 8.18 at a year.

Against the lower number the session is working harder than the lower number implies. The traded range on one publisher’s reading is 159.13 to 159.65, 52.0 pips, which is 0.898 of a one-week standard day and 0.768 of a one-month one. Both figures are rolling: that range is the extreme of a window still open, true only to that page’s 09:54 stamp. Monday ran 0.768 of a standard day and Tuesday ran 0.726 before its low extended and the true figure turned out to be 1.186. We are not going to call today’s number final.

The one-month 25-delta risk reversal reaches us at 1.6175 the negative way, labelled yen calls over, narrowed slightly. That is the fourth reading of this object in six days and the sign problem has not gone away: 1.6700 negative on 13 November, 1.68 positive on 17 November under the identical label, 1.6850 negative later the same day, and 1.6175 negative now. The magnitudes span 0.0675. The words disambiguate and the number does not, and a feed that stores the signed value without the label stores a coin flip.

And one long-standing absence partly closes. This desk has been looking for a dated 200-day moving average on dollar-yen for weeks and printing that it could not find one. A value now exists: 158.81, alongside a 100-day at 158.88 and a 21-day at 158.37, from one publisher. Spot in the 159.20s sits 44.0 pips above the 200-day, 0.760 of a standard day. The 100-day and the 200-day are 7.0 pips apart, which is inside a single hour’s ordinary travel and makes the pair useless as separate levels. And the page states no as-of date for any of the three. A moving average is a function of its window and its end date; without the date it is a number we can quote and cannot audit, so we are publishing it as exactly that rather than as a level.

What this does not tell you

  • Every notional in this piece rests on one publisher, reached through a mirror of its page because the original failed. Three attempts at an independent second reading of this cut returned three documents for other dates entirely — one for 18 June 2026, one for 26 May 2026, one for 13 November 2026. The June page carries dollar-yen strikes at 155.00, 161.00 and 158.00 with billion-dollar notionals that would look perfectly current to anything that did not check the byline. We publish the figures labelled, not corroborated.
  • Three observations of a blank cell is three observations. Two under a threshold and one 1.060 times over it does not establish a rate, and we are not quoting one.
  • The one-billion, two-billion and three-billion labels are the banded publisher’s own stated floors. Its sizes sit behind a paid membership, so we are testing a vocabulary against a different publisher’s numbers, not against the book.
  • The session high of 159.65 and the low of 159.13 are one publisher’s, stamped 09:54 on a page that prints no time zone for its own clock. The handle is corroborated: a second publisher’s index puts spot near 159.25 at 19:20 Tokyo time, which is 10:20 UTC, and that headline’s body did not load for us.
  • Our 15:00 UTC arithmetic uses 159.25 as the reference. Every pip distance in this piece moves with spot and none of it is a forecast.
  • The 158.50 strike appears on this cut at the notional publisher and on Thursday’s cut at the banded one. That is not necessarily a contradiction, because one strike can carry options expiring on several days — which is precisely why agreement on a strike cannot validate a cut date.
  • The 200-day at 158.81 is single-sourced and undated. It does not close the errand; it narrows it to asking one publisher for one date.

Related

  • Economic calendar and event windows — the 15:00 UTC cut is midnight in Tokyo, and the FOMC minutes four hours after it are 04:00 Thursday there.
  • Current signals — what a 57.88-pip standard day does to a stop that was sized on a 60.89-pip one.
  • Expert advisor presets — storing a size band and a distance as separate fields, and refusing to multiply them.

Sources read for this piece, all on 18 November 2026: SwingFish carrying the 18 November New York cut expiry list, bylined 18 November 2026 and attributed in its own body to a named writer at InvestingLive · Minkabu FX for the implied volatility term structure and the risk reversal, published 17:27 Tokyo time with its table marked as reference values at 16:34 · a market outlook on MQL5, stamped 18 November 2026 09:54 with no zone, for the traded range, the moving averages and the dollar index · Gaitame.com for the 158.800 to 160.000 forecast range and the prior session’s path.

Pages we tried and could not open, so that no figure here is implied to rest on them: the dated New York cut expiry page at Newsquawk, which has now failed at its dated slug for a fifth consecutive slot; the original InvestingLive page whose content we are reading through a mirror; and four Minkabu news items at identifiers 384837, 384833, 384790 and 384768, which is why the spot level in this piece is a headline-level reading. Three further expiry pages returned a clean response and the wrong date, and are named in the humility section rather than linked. The structured order board whose size bands this piece tests was not reached directly today; the bands quoted from it are this desk’s own carried readings of 17 and 18 November, and the one-billion, two-billion and three-billion floors are that page’s stated vocabulary.

All times are UTC unless a source’s own clock is named. Figures are as published by the sources above at the times stated; commentary and interpretation are our own. Nothing here is investment advice, a recommendation, or an offer to transact. Option notionals, implied volatilities and moving averages are observations about a market at a moment, not predictions about where it goes next.


Flow Desk
Flow Desk