New York closed the dollar–yen at 158.10. Tokyo’s low this morning was 158.10, to the pip. Tokyo’s high was 158.51, and the pair was rejected there and spent the rest of the session in the low 158.40s. Forty-one pips, one touch, no follow-through. This desk has a published resistance level at 158.50, which makes the high a one-pip overshoot and the session a clean score. It also has four separate explanations for why the pair stopped there, and no way whatsoever to choose between them. That is the useful part.
The session, in the numbers that are actually measurable
At 01:00 UTC the pair was 158.42, having traded 158.47 high and 158.10 low to that point. At 03:00 UTC it was 158.43, with the high extended to 158.51. The low never moved off the New York close. By roughly 05:10 UTC a second publisher had it clinging to gains near 158.50 and called it a one-and-a-half-week high; at 02:03 UTC the same publisher had it around 158.35 with a fourteen-day relative strength index of 56.27.
So the shape is unambiguous even where the attribution is not. The session’s entire range sat in the top forty-one pips of yesterday’s close, the downside was never tested, and the one attempt above 158.50 lasted long enough to print and not long enough to hold. Range character, with an upward drift and a ceiling.
What did not happen is worth stating plainly. The nearest published support on this desk’s three-day-old ladder is 157.50. The low was sixty pips above it. The next resistance above is 159.00, forty-nine pips above the high. The whole session took place inside one rung.
Four explanations for one pip, and not one of them has a number you can check
Why did it stop at 158.51? Take the candidates in order of how confident they sound.
One: this desk published 158.50 as resistance three days ago, and the high was 158.51. A one-pip overshoot of a named level is about as good as level analysis gets.
Two: a Japanese publisher attributes the rejection to the two-hundred-day moving average, saying the dollar climbed to 158.51, met resistance there and saw profit-taking back to around 158.30 before recovering.
Three: an English-language publisher also points at the two-hundred-day average — but describes it as a hurdle the bulls are still awaiting a move beyond, which places it above the current price rather than exactly at the high. The same page puts the hundred-day average at 159.55, a hundred and four pips above this morning’s high, and does not publish the two-hundred-day value at all.
Four: yesterday’s New York cut carried an expiry cluster at 158.00 to 158.05 worth slightly over one billion dollars. That is fifty-one pips below the high and five to ten pips below the close and the Tokyo low.
Now notice what you have. Two publishers naming the same moving average, one of them putting it at the high and the other above it, and neither printing the figure. A desk level that fits to a pip and was derived from a different publisher’s ladder three days ago. And an expiry that is demonstrably in the wrong place to have caused this.
The operational consequence is not “we do not know”. It is that these four things decay at different rates. An order level sits where it sits until it is filled or pulled. A moving average moves every single day, and on a pair that has climbed about a figure and a half in a week it moves upward fast. An expiry ceases to exist at 14:00 UTC. If you are carrying a short stop above 158.50 on the theory that it is an order cluster, and it was actually a moving average, your level has already drifted by tomorrow and you will not be told.
Today’s expiry ladder does not exist, and we are not going to pretend otherwise
The page that carries the New York cut expiries for 7 October returned a 404 when fetched at 06:14 UTC. The 6 October page serves, and it is where the 158.00 to 158.05 cluster comes from, but its notional figures are published inside an image rather than as text and the only number extractable as prose is the one quoted above. So the honest position is that we have yesterday’s ladder, not today’s, and yesterday’s ladder is about forty-five to fifty pips below where the pair is now trading.
The useful inference from yesterday’s cluster is a negative one, and negatives are cheap and reliable. A billion dollars of strikes at 158.00 to 158.05 did not stop the pair closing at 158.10 and it did not pull the Tokyo session back through it, despite the low sitting exactly at the close and therefore within ten pips of the top strike all session. If a one-billion-dollar cluster five pips away produces no visible pin, then the size at which expiries matter on this pair, in this liquidity, is larger than one billion. That is a sizing fact, and it survives not knowing today’s numbers.
Two data beats, two currencies, no bid in either
At 23:30 UTC Japan’s August labour earnings showed real wages up 1.5 per cent year on year, an eighth consecutive monthly gain, with nominal cash earnings up 3.8 per cent against a deflator the labour ministry held at 2.2 per cent. That is the single cleanest argument available for further tightening, published by the government, on a pair whose entire story this autumn is whether the central bank moves again. The yen weakened.
At 06:00 UTC Germany’s August industrial production printed up 2.0 per cent on the month against a consensus of 0.5 and a July reading of minus 1.1 — a three-point swing in the monthly rate and a one-and-a-half-point beat. The euro was near 1.1225, down 0.30 per cent on the day.
Two beats inside seven hours, two non-reactions. The yen one has a documented reason: October rate-rise odds are reported around 12 per cent, down from roughly 40 per cent a week ago, with December near 90, so a wage print that supports tightening is landing on a market that has already moved the tightening to a different meeting. The euro one has a reason too, and it is a better warning. The same German statistical office published August factory orders on 6 October at minus 10.6 per cent on the month against a consensus of minus 1 and a prior of plus 2.5, with the collapse widely attributed to large one-off orders unwinding.
So on the same month, from the same agency, in two consecutive sessions, a system keyed on surprise magnitude saw minus 9.6 and then plus 1.5 — opposite signs, and the first one six and a half times the size of the second. The euro is net lower across both. Whatever those two releases are measuring, it is not the same thing, and a filter that treats German August activity as one signal with two prints has been handed a contradiction rather than a confirmation.
Scoring the ladder, including the half that was never tested
The three-day-old ladder this desk has been carrying had resistance at 158.00, 158.40, 158.50 and 159.00, and support at 157.50, 157.00, 156.72 and 156.40 to 156.35. Through this morning: 158.00 and 158.40 both gave way and neither held on the way through. 158.50 was exceeded by one pip and rejected. 159.00 was never approached. The entire support side was never approached either, by sixty pips at the nearest rung.
That is three rungs with information and five without, and the honest reading of a ladder that scores three for eight is that the top of it was roughly right and the rest of it was decoration. The number that did work — 158.50 — is also the one we cannot distinguish from a moving average. We are not going to claim it as a vindication.
One further caution we are marking ourselves on. This desk published yesterday that the pair had moved 28 pips around an event we wrote about three times, and used that as a quiet-session benchmark. This morning gave 41 pips with no event at all — 1.46 times as much, on nothing. Twenty-eight pips was not a quiet day. It was a quiet day with an event in it, which is a different and more awkward fact.
What this does not tell you
We do not know what stopped the pair at 158.51 and this article does not conclude that anything in particular did. Four candidates, one fitting to a pip, two naming an indicator neither publisher quantifies, and one in the wrong place. A fifth possibility we cannot exclude is that nothing stopped it and the session simply ran out of buyers at the top of its range, which on forty-one pips of total movement is entirely plausible.
The Tokyo high and low are one publisher’s marks, read at two points in the session. A second publisher’s level at 05:10 UTC is consistent with them but is not an independent high. We did not obtain a second reading of the range itself, and on a forty-one pip session a six-pip disagreement between publishers — which is routine on this pair — would be fifteen per cent of it.
The real wage figures are a secondary reading: the issuing ministry’s release was not fetched and the publisher whose own report on it we tried returned a payment-required error, so the numbers come from a wire report at one aggregator and the identity does not quite close from the published rounded figures. Subtracting the 2.2 per cent deflator from 3.8 per cent nominal gives 1.6, not the 1.5 reported, and July gives 2.1 against a reported 2.0 — a consistent one tenth in the same direction in both months, which is what a rounded deflator does and is not evidence of an error.
The October and December rate-rise probabilities are one publisher’s figures, unnamed as to instrument, and this desk has separately carried 12 per cent from an overnight index swap reading. They agree, which is reassuring and is not the same as corroboration. And the two-hundred-day moving average value remains unpublished by anyone we read, which means the central ambiguity of this article is one number away from being settled and we could not find that number.
