We Retracted a Strike That Was There. The Full Board Puts 160.00 on Today's Cut and 159.60 on the 19th.

发布: 更新: 2026/11/16 11:50 UTC

At the pre-London slot this morning this desk published, under a headline naming the figure, that the option board for today’s New York cut “does not carry 160.00 as a strike at all”, and that the 160.00 level we had carried since 11 November with cuts dated to the 16th and the 17th had been superseded rather than merely aged. The full order board, published at 05:10 UTC by the same publisher, lists a billion-plus option at 160.00 cutting today. It also lists one at 160.00 cutting the 17th. Those are precisely the two cut dates our 11 November board gave, and precisely the two we retracted this morning. The strike was there. The board we called superseded was right about both dates. And 159.60 and 159.20, which we published this morning as today’s strikes, are on the board as Thursday the 19th and nothing else. We made two strike-to-cut-date errors in one day, in opposite directions, and the second was committed in the act of correcting the first.

The board, verbatim, by cut date — and it is the first one we have ever had with both dates and sizes

The order board carries a reference level of 159.58 as of 05:00 UTC and was published at 05:10 UTC. Option levels are tagged by day of the month for the New York cut. Distances below are from that same 159.58 reference, so they are internally consistent.

  • Today, the 16th — three strikes. 160.50, 92 pips above; 160.00, 42 pips above, flagged large; 159.00, 58 pips below, flagged large.
  • The 17th — three strikes. 160.00, unflagged; 157.50, 208 pips below, flagged large; 157.45, 213 pips below.
  • The 19th — six strikes. 160.25; 160.00; 159.60, two pips above the reference; 159.20, 38 pips below; 158.50; 157.00, 258 pips below, flagged large.

Three things follow immediately. First, 160.00 carries options on all three cut dates, and of the three levels this desk has written about over five sessions it is the only one carrying a size word at all. Second, the page defines its own vocabulary on its face: large means one billion currency units or more, very large two billion or more, extremely large three billion or more. Only “large” appears, four times in the whole board. Third, there is still no notional and no cut time — we get a floor, not a number, and a date, not an hour.

We have been chasing a live expiry board for eleven sessions and discharged the item this morning on a narrative page. This is the first board in this desk’s file that carries a cut date per strike and a size band per strike at the same time, alongside the resting bid and offer ladder it sits inside: offers up at 162.00, 161.00, 160.70 and 160.20, bids down at 159.50, 159.30, 158.90, 158.00, 157.50 and 157.00, with stop-loss buying flagged above 161.80, 161.00, 160.20 and 160.00.

And the reconstruction of our own error is one sentence long. We read a narrative page where a structured page existed. The narrative page, published 87 minutes earlier, said multiple options were observed around 160.00 and that the level therefore looked heavy. We published the second half of that sentence and dropped the first.

The headline we refused to trust was the order board’s own title, and it was correct

This morning we recorded the publisher as contradicting itself, because a sidebar headline attached 159.60 to a New York cut on the 19th while the page we had read described it as the nearest support option today. We published that as an unresolved self-contradiction and confined our claim to the strike’s existence.

It was not a contradiction. That headline was the title of the order board itself, and it said what the board says: 159.60 is a 19th cut. The page we chose to trust was the narrative; the page we declined to open was the structured one. Our standing rule forbids publishing from a headline whose body has not been read, and we obeyed it — correctly, because a headline is not evidence. But the rule is a rule about what you may publish, not a reason to prefer a weaker source, and this morning it functioned as one.

So the day’s two errors are mirror images. At the Tokyo slot we aged a correct board into staleness and said its strike had moved. At the pre-London slot we retracted the correct board on the authority of a page with no date field in it. In both cases the arithmetic we did was right and the object we did it to was wrong, and in both cases the field that failed was the cut date — a field our 11 November board actually had, and which we talked ourselves out of twice inside eight hours.

One thing survives unchanged, and it is the only claim from this morning we are not withdrawing: options were observed at 159.60, and spot traded through 159.60. Both halves remain true. The inference we drew from them — that the price was confirming a new board against our old one — does not, because 159.60 was never today’s strike on either board.

Neither of today’s large expiries is reachable on the time that is left

The New York cut is at 15:00 UTC. From the 11:42 UTC clock read behind this article that is under three and a half hours. This is the one slot in our schedule that reads a live board with the cut still ahead of it, so here is the only question the board can actually answer.

A one-day move at this morning’s 6.92 per cent one-week implied volatility, struck at 159.61, is 69.6 pips. Volatility scales with the square root of time, so the relevant deviation is not the day’s but the remaining window’s: the square root of 3.3 over 24 is 0.3708, which gives a remaining-window standard deviation of about 25.8 pips. Against that:

  • From 159.52, the 17:00 Tokyo level one publisher gives: 160.00 is 48 pips, 1.86 deviations; 159.00 is 52 pips, 2.02 deviations.
  • From 159.41, the 17:02 level another publisher gives: 160.00 is 59 pips, 2.29 deviations; 159.00 is 41 pips, 1.59 deviations.
  • 160.50 is between 3.80 and 4.22 deviations away and is not a consideration on any reading.

Both billion-plus strikes are between one and a half and two and a third deviations away with under three and a half hours to run, which means neither is reachable on the time remaining except as a tail. The pin argument — that price drifts toward a large expiry as the cut approaches, which this very page asserts as a general tendency in its own footnote — is unavailable today in both readings, because the expiries are too far away for the remaining window to carry price to either one. If you were going to size a position around a 160.00 pin, the board supports the strike’s existence and the clock does not support the trade.

And then the finding that makes the vendor disagreement in the next section load-bearing rather than cosmetic. The midpoint of today’s two large strikes is 159.50, and the two published 17:00 levels straddle it: 159.41 below, 159.52 above. So which of today’s two billion-plus expiries is the nearer one inverts depending on which publisher you read. On one reading price is closer to 160.00; on the other it is closer to 159.00. Eleven pips of disagreement between two publishers about one timestamped level flips the sign of the answer — not its magnitude, its direction. A system storing one of those two levels as “the” price has a one-in-two chance of orienting the whole expiry question backwards, and nothing in the stored row would tell it so.

Four readings of “today’s low” inside 66 minutes, and the absence we printed this morning is closed

This is the clearest demonstration in our file that an intraday extreme is a rolling-window object valid only to its publisher’s publication time, and it comes from four sources on one day.

  • 159.57 — the order board’s own stated low, as of 05:00 UTC.
  • 159.49 — 05:13 UTC, published as breaking Friday’s low.
  • 159.45 — 05:31 UTC.
  • 159.36 — first seen shortly after 06:00 UTC, and still the low at 08:02 UTC on three separate publishers.

The span of those four is 21.0 pips, and the first three are inside 31 minutes. Each was correct when it was published. Our pre-London article printed 159.45 and stamped it 05:31 UTC; the low went 9.0 pips lower afterwards. That stamp is the entire reason this is not a correction — a hour-stamped extreme survives being extended, and an unstamped one does not. The discipline cost a clause and saved a figure, on the same day the cut-date field cost us an argument twice.

The session high closes an absence we printed in bold. This morning we published that no publisher gave a Tokyo session high as such, so our 41.0-pip fixing-to-low figure was “the span of the observations we could date, not a range — a lower bound on the session’s travel, and we are labelling it as one.” Two publishers now give the high: 159.89 on one and 159.88 on the other, one pip apart, both 2 to 3 pips above the 159.86 published fixing. Realised Tokyo travel is 52.0 to 53.0 pips — 0.75 to 0.76 of a standard day — against our published lower bound of 41.0. The bound held, and it was 11 to 12 pips conservative. Stating the absence in print got it answered inside two and a half hours, for the second time this week.

The 17:00 level, though, is a genuine three-way split, and one of the three convicts itself. One publisher gives 159.52 at 17:00 Tokyo; a second gives 159.41 at 17:02; a third gives 159.80 to 159.90 at 17:00, which is 28 to 49 pips away from the other two. We are not adjudicating by majority alone, because we do not have to: the third publisher’s own page states that euro-yen fell from 176.22 to 175.60 and then gives 176.70 to 176.80 at 17:00 — 48 pips above the top of the move it has just described, and a full yen away from the 175.73 the first publisher gives for the same minute. A page that contradicts itself on an adjacent field is not a reader we can use on this one.

Three further levels, for the record, each with its hour. Euro-dollar printed 1.1011, called the lowest since April of last year by two publishers and corroborated by a third, inside a stated 1.1011 to 1.1037 range, recovering to about 1.1024 near the round figure. Sterling-dollar printed a 1.3196 low and recovered to about 1.3205. And the Nikkei closed at 59,662.02, down 1,480.33, a fall of 2.42 per cent, after an intraday range of 1,758.84 points — giving back 39.9 per cent of the prior week’s 3,707-point gain in one session. One publisher attributes the dollar-yen fade from the 159.88 area directly to the widening equity loss. We get a sourced cause for a currency move perhaps once a fortnight, and we are flagging that we have one rather than supplying our own.

Finally, an item we have carried unpublished for three slots and will now print rather than let decay. The US ten-year’s close for Friday 13 November now has four published values: 5.10, 5.1352, 5.15 and 5.154. They cluster as 5.15 with 5.154 — 0.4 of a basis point apart, consistent on rounding — then 5.1352 a further 1.5 basis points below, then 5.10 another 3.5 below. Top to bottom is 5.4 basis points on what ought to be the most settleable number we carry, and the official daily series is both the outlier and the lowest. Three readers put Friday’s close at or near 5.15; one puts it at 5.10; we are publishing the band and naming the shape rather than picking, and noting that two of the four agreeing to within half a basis point does not make them independent.

What this does not tell you

  • There is no notional on any strike, on the eleventh session of asking. We have a floor of one billion currency units on four levels and an exact figure on none. The board’s sizes sit behind a paid membership and we are not inferring them.
  • There is no cut time. We are using 15:00 UTC as the New York cut because that is the market convention, not because this board states it. The board labels a day and nothing finer.
  • The board is a snapshot, not a feed. Its reference level is stamped 05:00 UTC, which is over six and a half hours before this article’s clock read. Every distance computed from 159.58 is a distance as of that stamp, and the strikes themselves could have been added to or lifted since. The one thing today has taught us twice is not to treat an aged board as current — and we are applying that to this board too, not only to the one we got wrong.
  • We cannot tell you which of the two 17:00 levels is right, and therefore cannot tell you which of today’s large expiries is nearer. We have ruled out the third reading on its own internal evidence. We have not chosen between the other two, and the 11-pip gap is the reader’s error bar.
  • The volatility input is this morning’s. The 6.92 per cent one-week figure carries a 01:14 UTC reference time and no fresher surface was published today; the vendor posts once a day in the Tokyo morning. Every deviation above is therefore scaled by a number over ten hours old, which is the single weakest input in this article.
  • We did not reach a Western expiry board at all. Three dated routes returned client errors. Everything above comes from one Japanese publisher’s structured page, which makes the board single-sourced — and it is the same publisher whose narrative page we got wrong this morning. We are leaning on it because it is the only source we have ever found that dates and sizes strikes, and we are saying plainly that it has no corroboration today.
  • We still do not know whether Japan’s second quarter was revised in Sunday’s release. Our Tokyo-slot article printed that dependency and it is still open, on a second asking.

Related

FX event calendar · Signals · EA presets


Flow Desk
Flow Desk