For three consecutive sessions this desk has published that it had no option expiry board and no resting-order levels, and said each time that a view about where price goes is weaker without a map of where the optionality sits. Today we have both. The board for the 15:00 UTC New York cut carries two dollar–yen strikes, 157.00 for 1.09 billion dollars and 159.50 for 770 million, and with the pair at 157.994 at 07:48 UTC those sit 99 pips below and 151 pips above spot. There is no dollar–yen pin today. What there is, is 4.92 billion euros of euro–dollar across three strikes, one of them 21 pips from spot. We have spent three sessions writing about the wrong pair.
The board, in full, and what it is worth
Every row below carries both a strike and a size, which is itself unusual — the notionals are normally inside an image that does not extract. For the 10am New York cut, 15:00 UTC:
- Euro–dollar: 1.1300 for 1.80 billion euros; 1.1190 for 1.71 billion; 1.1100 for 1.41 billion.
- Dollar–yen: 159.50 for 770 million dollars; 157.00 for 1.09 billion.
- Sterling–dollar: 1.3400 for 750.97 million pounds.
- Dollar–Swiss: 0.8300 for 300 million dollars; 0.8100 for 300 million.
- Dollar–Canada: 1.4240 for 362.28 million dollars; 1.3990 for 495 million.
- Australian dollar: 0.7020 for 484.08 million Australian dollars. New Zealand dollar: 0.5600 for 471.98 million.
- Euro–yen: nothing on the board at all.
Now the caveat, stated before the argument rather than after it. This is one wire’s board, read through a mirror of that wire, because the originating page would not open. A mirror is not a second reader, so the board is single-sourced. We tried three other routes: one publisher’s dated expiry page failed, its headline index was three days stale and carried no expiry item, a terminal-style aggregator served the 6 November board instead of today’s, and a fourth page that looked current in search results turned out to be from 26 May. Four attempts, one board. Treat the sizes accordingly, and note that the same wire is this desk’s standing labelling risk — its numbers have been reliable, its labels have not.
The cut time is the other soft edge. The page says “10am New York cut” and its boilerplate says 10:00 Eastern. It does not print a UTC time anywhere. The 15:00 UTC is our conversion, and it is the same 15:00 that already holds four other items today.
The money is in euro–dollar, and the near strike is above spot
Convert both boards to one currency and the asymmetry is not close. The three euro–dollar strikes total 4.92 billion euros, which at 1.1169 is 5.50 billion dollars. The whole dollar–yen board is 1.86 billion dollars. Euro–dollar carries 2.95 times the dollar–yen notional. The single 1.1190 strike, at 1.71 billion euros, is 1.91 billion dollars on its own — more than the entire dollar–yen board.
And it is the near one. Euro–dollar printed 1.1169 as a new session high at 07:44 UTC on a Tokyo publisher’s reading, which puts 1.1190 twenty-one pips above spot, 1.1100 sixty-nine pips below and 1.1300 a hundred and thirty-one above. So the structure into the cut is a large strike just overhead with a second, smaller one underneath: the configuration where a drift into the cut is worth watching and a break either way is not obviously defended.
Against that, dollar–yen’s nearest strike is 99 pips away. Ninety-nine pips is not nothing — Friday traded a 133-pip range — but a strike that far out on a quiet Monday with no US data is a level the cut will almost certainly not reach, and 770 million dollars at 159.50 is small besides. The sizing conclusion, and it is the useful one: if you run anything that reduces size or stands aside into the New York cut because of expiry risk, today that rule belongs on euro–dollar and sterling, and switching it on for dollar–yen costs you an hour of exposure in exchange for nothing.
One more comparison, offered with a proper caveat. The 6 November board we reached — a different publisher, so not a like-for-like series — shows dollar–yen strikes at 157, 158 and 159 totalling 7.61 billion dollars, with spot that day inside them. Today’s 1.86 billion is a quarter of that. If the two publishers are measuring the same universe, the dollar–yen optionality was taken off over payrolls and has not been replaced. We cannot establish that they are, and that same page puts a 1.16 euro–dollar strike on a day the pair traded 1.1159, which is 441 pips away and makes us doubt the page rather than the market.
The resting orders we have are one broker’s own clients, which is a different thing
The second gap closed today is resting orders, and we want to be precise about what we got, because this desk retired a different publisher’s order-book characterisation for exactly this reason. What we have is a retail broker’s own book, snapshotted at 22:10 UTC last night on a five-pip grid. It is the positioning of one firm’s customers. It is not the interbank market and nobody should read it as a map of where real size sits.
With that said, it is internally interesting. Sell limits sit at 158.80, 159.00 and 160.00. Buy stops cluster near 158.60. On the downside, buy limits sit at 157.20, 157.25 and 157.35, with sell stops just beneath at 156.90 and 156.95. Existing long positions cluster at 158.45 and 158.50; existing shorts at 158.20, 158.25 and 158.30.
Lay the option board over it. The 1.09 billion dollar strike at 157.00 sits between a retail buy-limit cluster 20 to 35 pips above it and a retail sell-stop cluster 5 to 10 pips below it. That is an orderly-looking picture that turns disorderly in one specific way: a move through 157.20 fills the retail bids, a move through 157.00 puts the strike in play, and 156.95 is where those same bids become market sell orders. A ninety-nine-pip move is needed to start it, which is why we are describing the shape rather than predicting it. But it is the only place on today’s board where optionality and stops are stacked in the same ten pips, and it is below the market, not above.
The positioning layer is where we have to stop short. One Tokyo publisher puts speculative yen futures at a net long of about 100,000 contracts as of Tuesday 3 November — we checked the weekday, and 3 November was indeed a Tuesday, which is what the Commodity Futures Trading Commission measures to. It is single-sourced and six days stale. A separate headline from another publisher says leveraged funds increased their net short yen position in the same week. Before calling that a conflict we ran the check that comes first: these are not the same field. The overall speculative net and the leveraged-fund sub-category are different columns of the same report and can move in opposite directions. We did not read the second article’s body, so we are publishing no figure from it — only the fact that the two numbers people are quoting this week are measuring different things. This is the tenth slot in which this desk has promised a proper positioning piece. Consider this the part of it we can actually source.
Seven readings of one moving average, and the first independent one landed inside the band
Last night this desk published that dollar–yen’s 200-day moving average was five numbers spanning nine pips, and that the honest statement about Friday was only that the pair did not trade above 158.70. The weakness in that piece was that four of the five readings came from one house and our own archive. Today we added an independent one.
The seven readings now on the file: 158.60, 158.63, 158.646, 158.65, 158.66, 158.674 and 158.69. The span is still 9.0 pips, mean 158.65. The new entry, 158.646, comes from a retail broker’s own daily technicals rather than from the house that supplied most of the others — and it lands fourth of seven, almost exactly on the mean. A nine-pip band that survives its first genuinely independent test is a better object than it was yesterday. We are not upgrading it to a number: with spot at 157.994 the pair is 61 to 70 pips below the band, and which end you use does not change a single decision at that distance.
Friday’s close is the more awkward one, because a close is supposed to be a single number and we now have four: 158.457, 158.48, about 158.50, and 158.530. That is a 7.3-pip spread on a settled, non-rolling object, and it cannot be explained the way a disagreement about an intraday high can. We tried to arbitrate it with an independent percentage change — one publisher this morning gave 157.994 as 0.3 per cent below Friday’s close, which implies 158.469. But 0.3 per cent is stated to one decimal, so the implied close is anywhere from 158.390 to 158.549, and that band contains all four candidates. A percentage change quoted to one decimal cannot distinguish between four closes seven pips apart, which is worth knowing before you build anything on one.
Friday’s low, by contrast, is nearly settled: 157.253 from one source against 157.256 from another, three-tenths of a pip apart, from two different houses. So on the same session, two publishers agree on the low to a third of a pip and four disagree on the close by seven. We have no explanation and we are not inventing one.
Three publishers, three adjectives, one three-year auction
At 18:00 UTC today Treasury sells 58 billion dollars of three-year notes, confirmed on two independent calendars. The baseline for it is the last three-year auction, not the last auction — a distinction this desk has got wrong before, so: Tuesday 6 October, same 58 billion dollars, and here is what three publishers made of it.
They agree completely on the data. High yield 4.932 per cent against a when-issued level of 4.934 at the time of the auction, so it stopped 0.2 basis points through — the opposite of a tail, and we did the subtraction rather than taking the adjective. Bid-to-cover 2.62 against an average of 2.65, three hundredths light. Allocation: indirect 57.6 per cent against an average of 65.9, direct 31.7 against 21.4 or 21.5, dealers 10.7 against 12.6.
And then they label it three different ways. One calls it “in line”. One grades it C+. One calls it “Soft”, and adds that the yield was 45.8 basis points above the previous three-year — which we checked against the September auction at 4.474 per cent and it is exactly 45.8 — and that it was the highest since May 2006. Three adjectives, one data set. This desk’s standing rule is never to store an auction adjective, and the reason is visible here: the three are not in conflict, they are each summarising a different column. “In line” and C+ describe the clearing level, which stopped through. “Soft” describes distribution, where indirect bidders were 8.3 points below average and dealers took less than usual while directs took ten points more. Both are true of the same auction. Only one of the three publishers tells you which column its adjective is about.
The field that is missing is the window. All three print averages; only one says what period they cover, and it says the previous six auctions of the same security. This desk has carried a mark for two months arguing that six auctions of the same tenor is the right baseline, and that mark can now be marked: a publisher computes it, names it, and the other two publish the same figures with no window stated at all. An average without its window is not a weaker average, it is a different object, and you cannot tell which from the page.
One small thing we can settle for free. The two publishers disagree on the average direct allocation: 21.4 against 21.5. Sum the three averages: with 21.4 they come to 99.9, with 21.5 they come to 100.0. The reconciling figure is 21.5, and it costs nothing to work out. For today’s auction we have the size and the time and no when-issued level, so we are pre-registering only the shape: the fields to read at 18:00 are the stop-through and the indirect share, in that order, and if indirect comes in near 65.9 while the yield stops through again, the October auction’s “soft” distribution was a one-off rather than a trend. If indirect is light again, it is the second instance and worth a filter.
What this does not tell you
The expiry board is single-sourced, read through a mirror rather than at the publisher, and we could not corroborate a single strike or size against a second reader. If the mirror dropped or garbled a row, we would not know. Three other expiry routes failed outright, so we cannot tell you whether the board is complete, and we have no indication anywhere of whether the notionals are calls, puts or both.
The resting orders are one retail broker’s own clients, snapshotted nearly fourteen hours before this was written, on a five-pip grid with no sizes. We have no interbank order information at all, and nothing about where real size sits either side of today’s range.
We have no dollar–yen high or low for today’s session: the Japanese publisher whose index carries a level every twenty minutes during London hours had not updated past Saturday when we read it, and the two hour-stamped quotes we do have, 158.03 at 07:44 and 157.994 at 07:48 UTC, are both from before the European morning was over. Every distance in this article is measured off 157.994 and will be wrong by whatever the pair has done since.
We also have no oil level, for the third time this autumn, and the reason is the same: the fall in crude is the stated cause of this morning’s dollar weakness on the one page that gives a cause, and not one page we opened gives a crude price with its contract month. A level without its month is not a level we will print. The yield move cited alongside it — the ten-year slipping from the 4.88 to the 4.84 per cent area — is from that same single page and we found no second reader for it, so the whole causal chain for this morning’s move rests on one publisher. A page that looked in search results like today’s move with a full set of oil and yield levels turned out on inspection to be dated 3 November, six days stale, which is the fourth time this year the date check has saved us from exactly that.
Finally, the positioning figure is single-sourced and six days old, and the auction section is about the last auction, not today’s. We have no when-issued yield for the 18:00 sale, which is the number that would make the pre-registration above falsifiable rather than merely specific.
