Dollar–yen ran 158.95 to 158.05 on 14 October and closed at 158.33 — ninety pips of range and two pips of net change. Yesterday morning this desk published the option market’s own number for the day: one-week implied volatility at 6.17 per cent, about 61.5 pips of one-day standard deviation. The day traded 1.46 of those. And the level three separate readers had named, 159.04, held by nine pips. Both of those are scoreable, both were published in advance, and only one of them went our way.
Ninety pips against sixty-one priced: the option market was light by half
At 11:51 UTC yesterday we published a forward test. One vendor put one-week dollar–yen implied volatility at 6.17 per cent; run through a 252-business-day convention that is 61.5 pips of one-day standard deviation, and the Tokyo session had traded 40 pips, two-thirds of one sigma. The implicit question was whether the rest of the day would make up the difference.
It made up more than the difference. The full session ran 158.95 down to 158.05, which is 90 pips, or 1.46 sigma. The Tokyo portion we measured was 0.65 sigma; the fifty pips that arrived after it were worth another 0.81. So the day did not merely fill out its implied move, it overshot it by nearly half, and the overshoot was concentrated in the two hours around a single scheduled release.
If you size off implied volatility — and a system that uses an ATR-style denominator is doing something close to that — this is the uncomfortable case. The number was not wrong about the quiet part of the day. It was wrong about the whole day because the whole day was one event, and a one-day sigma spreads an event’s risk across twenty-four hours that did not share it. The practical correction is not a bigger sigma. It is a smaller position in the window and a normal one outside it.
Nine pips short of 159.04, and we should be honest about what that is worth
159.04 has been in this file for three sessions, named independently by three publishers as the level above the market where orders sit. Yesterday morning it was 69 pips away, 1.12 sigma. The high was 158.95. The market travelled 60 of the 69 pips — 87 per cent of the distance — and stopped nine pips short.
It is tempting to read that as the level working. Resist it. A nine-pip shortfall on a ninety-pip day is inside the noise of the measurement itself: two of our three sources for the high quote it to the sen, the third rounds, and the difference between a high of 158.95 and one of 159.01 is a single print on a thin book in the minutes after a print. What yesterday establishes is weaker and more useful than “159.04 held”: a session that overshot its implied move by 46 per cent, on the hottest inflation surprise of the quarter, still could not reach the level. That is a statement about how much fuel the move had, not about the integrity of a round number.
The level is now 71 pips above this morning’s open, which is further away than it was yesterday, on a day with no comparable scheduled catalyst until 12:30 UTC.
Three readers, one high, and the same desk disagreeing with itself about the close
The high is as well sourced as anything in this file: three publishers give 158.95, one of them calling it the highest since 24 September. The close is not. One surface carrying the Fisco desk’s New York wrap says the pair fell back to 158.05 and closed there. A second surface carrying the same desk’s note gives the low as 158.05 and the close as 158.33. A third publisher’s 07:00 JST quote table has 158.32 to 158.33 and a day-on-day change of plus two pips.
Two of three agree, and the arithmetic settles it: a close at 158.05 against the previous 158.34 is not a plus-two-pip day. So 158.33 is the close and 158.05 is the low, and one publishing surface has collapsed a low into a close. We are logging it as the third instance of a single publisher running two versions of the same number on two pages, because it is now a pattern rather than a slip, and because this time the two versions differ by 28 pips on the field a reader is most likely to anchor on.
One publisher says the CPI lifted it above the 200-day. Another says it failed to close above. Both are right.
A research note published during the New York morning carried the headline that hot US inflation had lifted the pair above its 200-day average. At 22:33 UTC a Japanese broker’s morning piece said the pair had failed to close above the 200-day. Neither printed a value, and the first does not state a timezone on its own byline — it gives 3:17 p.m. and leaves you to guess, which is the seventh time this file has recorded a published market figure with no readable as-of time.
This desk carries two readings of that average because it has spent two weeks establishing that there are two: a simple band at 158.50 to 158.56, and our own exponential reading just under 158.00. Against the simple band, the 158.95 high was 39 pips above it and the 158.33 close was 23 pips below it — so the pair went above the average intraday and failed to close above it, which is exactly both headlines at once. Against the exponential reading, the close is 35 pips above and nothing happened at all.
So the two publishers are not disagreeing about the market. One is describing the afternoon and the other is describing the close, and a reader holding only one convention cannot tell. This is the sixth unlabelled moving average in this file and the second time in a week that an unstated convention has decided whether a published event occurred. If your system has a 200-day filter in it, the convention is not a detail; yesterday it was the difference between a break and a non-event.
Tokyo opens flat, and the one forecast we can read has narrowed by 120 pips
The pair is 158.32 to 158.33 at 22:00 UTC, two pips above where it started 14 October. The entire ninety-pip excursion is gone from the price and remains only in the range.
The one piece of forward furniture available this morning is a broker’s forecast band of 157.50 to 158.80, which is 130 pips wide. The same publisher’s evening band yesterday was 157.00 to 159.50, or 250 pips. We do not trade these levels and we have said so repeatedly, but the width is information the levels are not: the publisher that was willing to hold open a 250-pip day in front of the print has shrunk to 130 pips behind it, and has moved its upper bound below the high the market actually printed. Read as a confidence statement rather than as support and resistance, that is a publisher saying the event risk is spent.
The calendar does not entirely agree. US PPI, retail sales, claims and two regional manufacturing surveys all land in the same minute at 12:30 UTC, and one of the two Fed officials who spoke yesterday named that PPI release by name as the input that will inform his inflation outlook. A five-release minute, the day after a print that moved the October meeting by twenty-four points, is not a day to inherit yesterday’s narrower band.
What this does not tell you
The 6.17 per cent implied volatility that the whole sigma argument rests on has one reader and no second publisher in two sweeps has quoted an implied volatility at all. If 6.17 is wrong, every pip count in this article scales with it, including the 1.46. The 61.5-pip figure is also ours, not the vendor’s: it is 6.17 per cent scaled by a 252-business-day convention at a spot of 158.35, and a calendar-day convention gives a different weekly number.
The low of 158.05 has one reader. The high has three and the close has two. We have no intraday path at all — we know the day ran 90 pips, not whether it ran them in one move or three, and the sigma comparison would mean something different in each case.
No crude level was published anywhere we could read, on a session whose own wrap attributes the oil move to a Houthi attack on Saudi Aramco. That is the thirteenth consecutive session in which an energy driver sits in the feed and this desk has no level to put next to it. We are not going to invent one, and we are not going to assert that oil did or did not contribute to a yen move we cannot decompose.
Nothing here is a view on direction. The pair is two pips from where it was and we have no idea which way it goes.
Related
- Economic calendar and event windows — the 12:30 UTC five-release minute is on it.
- Signals — for the range-versus-trend character of a session like this one.
- Expert advisor presets — where a volatility denominator and a news filter actually get set.
- The Option Market Priced a Sixty-One-Pip Day — the article this one scores.
- Two New Readings of the 200-Day — the convention problem, set out.
