At 06:16 UTC yesterday this desk published a structural case for why dollar–yen could not get up: with spot at 157.92 there were four named levels inside 64 pips above it — the 21-day at 158.02, the 10-day at 158.23, the Ichimoku conversion at 158.25 and the 200-day at 158.56 — in a market averaging 68 pips a day. Twenty-four hours later the pair is at 157.03 on one reader’s 05:20 snapshot and the nearest resistance anybody names is the daily Ichimoku conversion line at 157.82. That is 79.0 pips away. The first level above spot today is further than yesterday’s entire four-level stack. Below, one publisher names support at 156.06 and then nothing until 155.34, the 17 September low: a 72.0-pip void underneath the only support in view. The ladder did not loosen. It inverted.
Yesterday four levels sat inside 64 pips. Today the first one is 79.
Take the two mornings as a pair, because they were measured the same way from the same publisher’s daily technical note.
- 22 October, 06:16 UTC. Spot 157.92. Four levels between 158.02 and 158.56 — a 54-pip band, all of it inside 64 pips of spot. Nothing named below until the Ichimoku base at 157.19, 73 pips down.
- 23 October, 05:20 UTC. Spot 157.03. Resistance 1 at 157.82, with the base line in the 157.40s between. Support 1 at 156.06, support 2 at 155.34.
The operational difference is not that the numbers changed. It is that the density moved sides. Yesterday a rally had to clear four prices in 64 pips and a sell-off had nothing to lean on; today a rally has 37.0 pips of air to the base line and another 42.0 to the conversion line, while a sell-off meets one price at 156.06 and then falls through 72.0 pips of unnamed space.
If you size a mean-reversion short off a named resistance, yesterday gave you four entries inside a 54-pip band and today gives you one at 157.82. If you size a breakdown continuation, yesterday gave you a 73-pip run to the first support and today gives you 97.0 pips to 156.06 and then a 72.0-pip gap. Same instrument, same publisher, same indicator set, opposite exposure profile, one session apart. A structural case has a shelf life and this one’s was a day.
A cloud bottom is not a level. It is a level with a date on it.
This matters because of a number this desk wrote down yesterday and was about to publish as a measurement dispute. One publisher’s 22 October indicator page put the daily Ichimoku cloud lower bound at 155.65. The publisher we read this morning puts it at 156.06, and its own earlier note today put it at about 156.05 — internally consistent to a single pip. Our note said: 41.0 pips apart on the only named support below spot, two publishers, and the pair is within sixty-odd pips of the higher one. That looked like exactly the kind of vendor disagreement this desk goes looking for.
It probably is not one. The lower bound of an Ichimoku cloud is not a constant anybody chooses; it is the leading span, plotted 26 sessions ahead of the prices that generate it. It takes a new value every day by construction. Comparing a 22 October reading with a 23 October reading and calling the difference a disagreement is a category error, and it is ours, not the publishers’.
So we are retracting the dispute before publishing it, and the correction generalises: every level this desk carries needs an as-of date, and the ones computed from a rolling window need it most. A 200-day moving average, a cloud bound, a Bollinger band, a pivot — each of these is a function of a window, and two readings of it taken a day apart are two different objects that happen to share a name. Only levels anchored to a single event survive comparison without a date: the 17 September low at 155.34 is the same number next week. Support 1 is not.
We cannot close the 41.0-pip question properly today, because the publisher carrying 155.65 did not open for us — four attempts on four of its pages, every one a client error. That is stated here rather than resolved.
Tokyo took 29 more pips off the low and then gave 63 of them back
The session itself was small and it ran the right way for the inversion above. One publisher’s 00:04 UTC Tokyo note has the high at 156.81 and the low at 156.40 — a 41.0-pip range — with the pair in the 156.50s at the time of writing. The 156.40 is 29.0 pips through the 156.69 that yesterday’s low is put at, and it is the lowest print anybody in this sweep names since 30 September.
By 05:20 UTC a different reader has the pair at 157.03, which is 63.0 pips off that low and 26.0 pips above the 156.77 the first publisher calls yesterday’s close. So the new low lasted about five hours and the recovery is bigger than the range that produced it.
The useful number in that sequence is the distance the low stopped at: 156.40 is 34.0 pips above the 156.06 support, which means the only named level below spot has not been tested yet. Everything above about the 72.0-pip void is a statement about what happens after a test that has not happened. Treat it as a sizing constraint for the London session rather than a prediction of one.
The expiry board did not open, for the fourth slot running the finding is about us
Yesterday this desk established the publication cadence for the option-expiry board we score each day: 05:51:19 UTC on 22 October and 05:10:57 on 21 October, a lead of roughly eight to nine hours on its own 14:00 cut. We also established that a search run at 06:11 could not find a document that had existed since 05:51 — a 20-minute index lag in our own retrieval.
So this morning we went straight at the URL pattern instead of searching for it, which is what the cadence finding was for. Three attempts, three client errors, nothing retrieved. The board may well exist; we could not read it, and under this desk’s own rule a document you cannot read in full is not a document you publish from. There is no expiry analysis in this note and the reason is retrieval, not absence.
That is the fourth consecutive slot in which the expiry question has resolved into a statement about our pipeline rather than about the market. It is worth saying plainly what that means for a schedule: a 06:00 entry that depends on a specific document being fetchable at 06:00 has a failure mode that has now fired on most of the days we have measured, and a schedule entry whose precondition fails most of the time is a schedule entry that should carry a stated fallback. Ours today was the level ladder above.
What this does not tell you
The level set in this note is one publisher’s. Support 1, support 2, the base line and the conversion line all come from the same daily technical page, and we have no second reader on any of them today. If that publisher’s cloud is wrong, the 72.0-pip void is wrong and the 97.0-pip distance is wrong with it.
The spot levels are three different publishers at three different moments — 00:04, a stated 05:20 snapshot, and a previous-day close — and none of them is a closing print on an instrument that does not close. The 63.0-pip recovery is the difference between two readings, not a measured path, and we have no intraday series for the interval.
We have no option-expiry board, no notionals and no strike distribution for today’s 14:00 cut. We have not established whether the 155.65 reading is wrong, stale or simply a different day’s object, only that it cannot be compared with today’s figure as if it were the same quantity. And nothing here is a view on direction: the whole argument is that the two sides of spot now carry different amounts of structure, which changes where stops are cheap and where they are expensive, and says nothing about which way the pair goes.
Related
- FX events calendar — the 14:00 UTC option cut and the European flash PMI cluster, in UTC.
- Signals — how a level ladder with density above differs from one with density below.
- EA presets — stop placement when the nearest named level is 79 pips away.
