On Friday morning this desk published a one-week dollar-yen implied volatility of 10.00 per cent and argued that the curve was inverted because the risk was in the next few sessions. This morning the one-week is 8.61 per cent. It fell 1.39 volatility points over the weekend — 7.3 times as far as the one-month, which fell 0.19, and 13.2 times as far as the average of the back end, which fell 0.105. The curve is no longer inverted. The one-week is now the lowest point on it and the one-month is the highest. The event premium was priced to a date, the date passed, and the premium went to zero on schedule. This is the cleanest confirmation of a published structural claim this desk has had, and the useful part is what the new shape says about the next four weeks rather than the one that just ended.
The whole curve, before and after
Dollar-yen implied volatility, same publisher, same page series, read at 08:12 on 2 October and at 02:10 on 5 October:
- One week: 10.00 to 8.61, down 1.39
- One month: 9.45 to 9.26, down 0.19
- Three months: 8.92 to 8.80, down 0.12
- Six months: 8.95 to 8.81, down 0.14
- Nine months: 8.92 to 8.81, down 0.11
- One year: 8.86 to 8.81, down 0.05
On Friday the one-week stood 1.14 points above the one-year. This morning it stands 0.20 points below it. The three-, six-, nine- and twelve-month tenors have collapsed into a 0.01-point band at 8.80 and 8.81, which is as close to a flat line as this surface gets.
What is left is not a curve at all but a single hump at one month, standing 0.65 points above the one-week and 0.45 above the one-year. The option market has one thing it is worried about, and it is between one week and one month away.
It is not hard to say what. Inside that window sit US September CPI on 14 October, the FOMC on 27 and 28 October, the Bank of Japan on 29 and 30 October, and the US midterm elections on 3 November. Outside the one-week window there is nothing comparable. A one-month option is the cheapest instrument that contains all four, and the shape of this curve says that is what is being bought.
What the levels mean in pips, and why the Tokyo range was not a surprise
At 8.61 per cent and a spot of 158.16, a one-standard-deviation day is about 86 pips. On Friday morning, at the one-month 9.16 reading we published then, the same calculation gave 91. At the 10.00 one-week it gave 100.
The Tokyo session ran 157.60 to 158.17, a range of 57 pips. That is 0.66 of a one-standard-deviation day at the new one-week level. A 57-pip range on a day the option market prices at 86 is an ordinary session, not a quiet one — and anyone sizing off Friday’s 100-pip figure this morning is carrying roughly 14 per cent less position than the surface now justifies for the same risk budget. That is the practical content of the vol collapse: it is a position-sizing input that changed by a seventh over a weekend, and nothing about the price action advertises it.
The one-week number also puts a figure on the week itself. At 8.61 per cent, one standard deviation over five trading days is about 192 pips, against 223 at Friday’s level. If you are trading a range strategy against the 156 to 158 box this desk has been describing, the option market is telling you the box is roughly a one-sigma week wide. That is not a reason to expect it to hold. It is a reason to know that it holding and it breaking are both ordinary outcomes, and to stop treating the break as the informative event.
The skew came in while spot made a new high, which is the interesting part
The 25-delta one-month risk reversal, same publisher: 2.30 to 2.67 on 2 October, yen calls over. This morning, 2.07 to 2.44, yen calls over. The midpoint fell from 2.485 to 2.255, down 0.23 of a point. The bid-ask width is identical at 0.37 on both readings, which is a small piece of evidence that this is a real move in the mid rather than a quoting artefact.
So the premium for yen upside protection got cheaper by about a quarter of a volatility point, over a weekend in which dollar-yen rose and then made a new Tokyo high at 158.17. Those two facts belong together. They do not add up to a bullish signal on the dollar. They add up to this: the market has stopped paying as much for the specific scenario in which the yen rallies hard, and it has done so at a spot level 26 pips above Friday’s New York high and 34 pips below the 200-day moving average at 158.51.
The skew is still positive, and that matters more than the direction of the change. Yen calls have been over at every reading this desk has taken, through a 25-basis-point Bank of Japan hike, a coordinated intervention and a payrolls shock. Nobody is selling yen upside cheaply. What has happened is that they are selling it slightly less expensively, which is a long way from the other side.
One standing caution, because it has bitten parsers before: the two publishers this desk reads write this quantity with opposite signs. One writes the yen-call premium positive, the other negative, for the same number. Our other reader’s most recent comparable figure was plus 2.47 on 1 October, against the 2.255 midpoint here — a fall of 0.215 across two publishers and four days, which is consistent with the same-publisher move of 0.23 but is not independent evidence of it.
We guessed at why two readers disagreed about Friday’s high. The same publisher carries both answers.
On Saturday this desk published two incompatible readings of Friday’s dollar-yen range: 156.95 to 157.91, a 96-pip day, from two readers; and 156.95 to 158.22, a 127-pip day, from a third. We offered a reconciliation and were careful to label it a guess — that 158.22 might be a full-day figure and 157.91 a New York-session figure, so the two pages were measuring different windows rather than disagreeing.
That guess is now settled, and it was settled by a publisher we were already reading. Minkabu’s Tokyo pivot page, timestamped 23:11 on 4 October, gives Friday as close 157.85, high 158.22, low 156.95 — the wide reading. Minkabu’s own New York recap gives the 156.95 low and a recovery into the 157.90s — the narrow one. One publisher, two pages, both windows, no contradiction. The 31 pips between the two highs is the gap between the full day and the New York session, exactly as hypothesised.
Two things follow, one about the market and one about us. About the market: the 158.22 full-day high is the relevant figure for anything that keys on daily extremes, and the 157.91 New York figure is the relevant one for a strategy that trades the US session. Using the wrong one costs 31 pips of stop placement. About us: the same-publisher check — look at the publisher’s own index for a second page covering the same date — is a standing rule on this desk, and we did not run it on the range before publishing two readings as a dispute. Running it took one fetch.
The pivot page also gives today’s levels off that wide range: resistance at 158.40 and 158.94, pivot 157.67, support at 157.13 and 156.40. Spot came within 23 pips of the first resistance in Tokyo without touching it. London has the busiest liquidity of the day and two obstacles stacked 11 pips apart — the 158.40 pivot resistance and the 158.51 two-hundred-day average. That is where this gets tested, and the option market is pricing the test as an ordinary day.
What this does not tell you
The volatility figures are one publisher at two timestamps. That is the right comparison for a change — same page, same construction — and it is the wrong comparison for a level, because we have no second reader for this morning’s 8.61. Our other option source publishes a one-month but not a one-week, so the single most important number in this article is single-sourced and will stay that way until someone else prints a one-week tenor.
We also cannot tell you that the one-month hump is about CPI, the FOMC, the Bank of Japan or the midterms. We can tell you all four are inside the window and nothing comparable is inside the one-week. Which of them the premium is actually for is not readable from a term structure, and a reader who concludes from this that the FOMC is the trade is doing the inferring, not us.
The implied-move arithmetic uses 252 trading days and a spot of 158.16 and is ours. It is a convention, not a market quote; a desk using 260 days or a different spot reference gets a slightly different pip figure from the same volatility, and the difference is a pip or two, not a sizing decision.
Finally, the skew reading and the volatility reading are from pages timestamped about eighty minutes apart this morning, and spot moved during the Tokyo session afterwards. Both predate the 158.17 high. Treat them as the morning’s surface, not as live quotes going into London.
Related
- FX events calendar — the releases inside the one-month window, in UTC
- Signals — volatility-scaled sizing and range character
- EA presets — range-versus-trend filters and expiry handling
