Dollar–yen traded 158.35 down to 157.55 on 15 October. Four publishers give that low to the pip and not one of them disagrees. The close is another matter: one reader puts it at 158.00, two others at 158.24 and 158.25. The 158.00 is the option strike this desk published yesterday morning without a notional and described, in print, as a location rather than a force. So the single number that would score our own forward claim is the one number in the session the published record cannot agree on — and at 07:00 JST this morning Tokyo is trading 158.01 to 158.02, which settles nothing at all.
The low is the most agreed figure of the week. The close is the least.
157.55 appears at four separate publishers: a Japanese four-values table stamped 06:30 JST, two Fisco New York wraps stamped 03:42 and 04:24 JST, and a Tokyo-outlook note stamped 08:00 JST. Same figure, to the pip, across three publishing houses and five hours of stamps. That is as clean as this file has ever seen a session level.
Now the close. The four-values table gives 158.00. The 04:24 Fisco wrap says the pair rebounded to 158.25 “and closed there”. The 03:42 Fisco wrap — same publisher, forty-two minutes earlier — gives the rebound as 158.24 and does not call it a close. The 08:00 outlook gives the rebound as about 158.25 and declines to state a close at all.
That is 25 pips between the highest and lowest account of where one session ended, with one pip of internal disagreement inside a single publisher. Note which reader produced the outlier: the one that published a structured open–high–low–close table. The two that put the close 24 and 25 pips higher are prose wraps describing a rebound, and a rebound is not a close. We think the table is more likely right about the field it was built to carry. We are not going to pretend that is the same as knowing.
We published a strike at 158.00 with no notional, and we still cannot score it
Yesterday morning this desk published that the 15 October New York cut carried a dollar–yen strike at 158.00, about five pips from spot at 05:56 UTC, and that the notional amounts existed only inside an image file. The published sentence was that a strike without a notional tells you where the gravity is and nothing about its mass.
One reader now says the day closed on that figure exactly. Three reasons that is not evidence of a pin, and we would rather state them than collect the coincidence.
First, the notional was never published, so there is no quantity against which the outcome could be scored. A strike that attracts spot and a strike spot happens to end near are indistinguishable without a size. Second, two readers put the close 24 and 25 pips away from it, which means the observation itself is contested. Third, 158.00 is a round number, and prices end near round numbers for reasons that have nothing to do with anybody’s expiry book. A forward claim you can only score against one of three published closes was not a falsifiable claim. That is a fault in how we framed it, not in the data.
Eighty pips, and the entire upside was in the first two
The four-values table gives the open at 158.33 — which matches the 14 October close this desk carried, a continuity check the table passes — and the high at 158.35. Two pips above the open, for the whole session. The low is 78 pips below it. Net change from open to the table’s close is minus 33.
So the shape is not a range and it is not a trend. It is a session that spent its entire budget in one direction, failed to hold the extreme, and gave back 45 of the 78. On the single-source one-week implied volatility this desk has been carrying — 6.17 per cent, one reader, never second-sourced — 80 pips is about 1.30 of a day’s standard deviation, against 1.46 the day before. Two consecutive days above one sigma, and the second one was quieter than the first despite six American releases landing inside it.
If your filter sizes from realised range, it is now reading two large days in a row and will size the third one up. The second of those two days had the larger event load and the smaller move. That is worth knowing before the filter does the arithmetic for you.
What Tokyo inherits: a reference level published four ways and two policymakers inside the session
The 200-day moving average is the level everybody is watching and nobody has one value for. This morning’s two Gaitame notes give it as about 158.57 and about 158.58 — the same publisher, seven minutes apart, one pip different. This desk already carries 158.50 and 158.56 from earlier in the week. Four readings, an 8-pip band, on a level being used as the session’s cap. Spot at 158.01 is 49 to 57 pips below it depending on whose average you take, and the direction of the answer does not change, which is the only reason the band is tolerable.
The events are the unusual part. Fed chair Warsh is in dialogue with the IMF managing director at 03:30 UTC — 12:30 JST — and one reader calls it his last public remarks before the FOMC blackout. The BOJ governor’s speech to the credit cooperatives convention lands at 06:30 UTC, 15:30 JST, and is being read by the deputy governor rather than delivered. Both sit inside Tokyo and early-London hours. A Fed chair speaking into Asian liquidity four hours from now, with no American desk fully staffed, is a different risk from the same words at 14:00 UTC.
Underneath that, one reader reports the Japanese finance minister saying after the G20 meeting in Bangkok that he is watching the market with a high sense of tension, that the G20 reaffirmed its currency commitments, and that Japan would act without hesitation if necessary. One reader, Japanese-language, from a publisher this file has found weak on proper nouns. We are reporting the remark and not building on it.
What to actually do with this
Three things, none of them directional.
Do not record 158.00 as yesterday’s close in anything that later scores a signal, unless you also record which publisher said so. If your backtest ingests one vendor’s daily close, it has already silently chosen between 158.00 and 158.25, and a 25-pip difference on a one-day bar is a large fraction of the thing most overnight filters measure.
Do not treat the 200-day as a line. Treat it as 158.50 to 158.58 and ask whether your rule survives the worst eight pips of it. If it does not, the rule is not about the moving average.
And put the two speeches in the schedule as Tokyo-hours events rather than Fed events, because that is when they will arrive and that is the book that will have to absorb them.
What this does not tell you
It does not tell you which close is right. We have argued the four-values table is the better-built field and that is an argument about construction, not a measurement. If the 158.25 is correct, the session gave back only 53 of its 78 pips rather than 45, the net change is minus 8 rather than minus 33, and the expiry coincidence disappears entirely.
It does not tell you the notional at 158.00, which has never been published in text and is not knowable from here. Nothing in this piece should be read as a claim that the strike moved spot.
The implied volatility benchmark is one reader and eight days old in its sourcing; the sigma figures are arithmetic on a number this desk has failed to second-source for four slots, and they should be read as a scale rather than a level.
The finance minister’s remarks are single-sourced in translation. The G20 characterisation in particular is the kind of claim that moves 40 pips if a wire confirms it and nothing if it does not.
And none of this is a view on where dollar–yen goes. A session that spent 80 pips to finish within 33 of its open, in front of a Fed chair and a central bank governor inside the same eight hours, is a sizing problem rather than a direction problem.
Related
- FX event calendar — the Warsh and Ueda windows in UTC
- Signals — range-based sizing after two consecutive one-sigma days
- EA presets — news-window handling for Asia-hours policy speeches
- Three positioning numbers before the 12:30 block, all broken differently — where the 158.00 strike was published without a notional
- Ninety pips against sixty-one priced — the previous session, and the implied volatility benchmark used here
