Overnight, one-month dollar-yen implied volatility went from 8.86 per cent to 9.16 per cent. The three-month went from 8.85 to 8.91. The front end moved five times as far as the tenor just behind it, and at 9.16 against a spot of 157.61 a one-standard-deviation day is about 91 pips — which is 2.12 times the entire Tokyo range of yesterday’s session. The non-farm payrolls report lands at 12:30 UTC. The option market has already told you how big to expect it to be. It has also, in the same overnight session, reduced what it will pay for protection against the yen going the other way, at every single tenor.
The front end repriced and the curve behind it barely noticed
Fisco’s currency-option note, stamped 02:41 JST, gives four tenors of dollar-yen implied volatility and all four of them higher. One month 8.86 to 9.16. Three month 8.85 to 8.91. Six month 8.75 to 8.84. One year 8.75 to 8.84. Read the changes rather than the levels: plus 0.30 of a volatility point at one month, plus 0.06 at three months, plus 0.09 at six months and at a year.
That shape is not a view about the yen. A view about the yen lifts the whole curve. What lifted here is the single tenor that contains today’s payrolls print, the 27–28 October Federal Reserve meeting and the 29–30 October Bank of Japan meeting, and nothing beyond it moved more than a tenth of a point. The market is buying a calendar, not a direction. If your volatility filter reads a single tenor and treats a rise as risk-off, it has just mislabelled a date.
The skew moved the other way, which is the part worth your attention
In the same note, the 25-delta risk reversal — the premium of yen calls over yen puts — fell at every tenor. One month plus 2.52 to plus 2.47. Three month plus 2.07 to plus 1.96. Six month plus 1.47 to plus 1.40. One year plus 0.64 to plus 0.61. The publisher’s own characterisation is that yen put buying, on an expectation of a weaker yen, strengthened relative to yen call buying for downside hedging. That is one desk’s reading of why; the four numbers are the fact.
Put the two together and the position is specific. More is being paid for a large move. Less is being paid for that move being yen-positive. Those are not contradictory and they are not a forecast either: they describe a book that wants gamma into the print and does not want to pay up for the crash hedge. The honest translation is that the people who price this for a living think today is wide and skewed slightly against the yen, and they have put a number on the width and only a few hundredths of a point on the direction.
Ninety-one pips is a sizing instruction, not a forecast
Convert 9.16 per cent annualised into a daily number — divide by the square root of 252 — and you get 0.577 per cent, or about 91 pips at 157.61. One standard deviation. Which means roughly a one-in-three chance the day finishes outside that band, in one direction or the other.
Now set that against what the last two sessions actually did. Yesterday’s overnight range was 123 pips on this desk’s own record. The Tokyo session that followed was 43 pips. Today’s implied daily move is 2.12 times that Tokyo range and about three quarters of the overnight one. If you carried position sizing calibrated on a 43-pip session into a 91-pip one, your stop distance is wrong by a factor of two and your risk per trade is wrong by the same factor. That is the only actionable thing in this article, and it does not require you to have any opinion at all about whether payrolls beats.
The number everyone is sizing against is three times the trend
The consensus for September payrolls is about 90,000 at FXStreet, 89,000 in one calendar’s own forecast cell, around 100,000 at another publisher, and around 60,000 at one named bank. Against that, the Bureau of Labor Statistics’ own August release states the average monthly gain over the prior twelve months as 31,000. The central forecast is 2.9 times the trailing trend of the series it is forecasting.
And the error bar on that forecast is not small. Taking one calendar’s own table of actual against forecast for the last four completed months — August 162,000 against 55,000, July minus 23,000 against 85,000, June 57,000 against 114,000, May 172,000 against 85,000 — the mean absolute miss is 89,750 jobs. This desk published 89,800 for that same quantity yesterday morning, from a different source. Two independent tables, fifty jobs apart. For once a figure of ours survives a second reader cleanly, and what it says is that the forecast and its own average error are the same size.
So the distribution you are sizing for is roughly nothing to 180,000, centred on a number that is three times trend. Ninety-one pips of implied daily range is not the option market being dramatic.
Where the levels are, and the positioning number that lands after the close
Spot was 157.61 at 08:57 UTC with the dollar index at 101.88 against a stated yearly high of 102.20. Tokyo closed near 157.54, about 71 pips below the New York close this desk recorded at 158.25. The 20-day exponential moving average sits at 157.26, the 30 September low at 156.38, the 24 September high at 159.04 and the 2 September high at 160.39, with the 14-day relative strength index at 51.50 — which is to say, nothing. Yesterday’s pivot set from a second publisher puts the pivot at 157.91, support at 157.37 and 156.80, resistance at 158.48 and 159.02. Spot entered the European afternoon 30 pips below the pivot and 24 above the first support. The 24 September high and that second resistance are two pips apart, which is a coincidence and not a confluence, but it does mean two unrelated readers nominate the same handful of pips.
Then at 19:30 UTC the Commodity Futures Trading Commission publishes the Commitments of Traders report, measured at Tuesday’s close. This desk published yesterday that the net yen position’s median week-on-week change over the last eight releases is 95,000 contracts against an eight-week range of 284,000, with three sign flips in eight transitions, and argued that the three-day lag matters less than the instability. Tonight is the first test of that claim. It is also measured before today’s print, which means whatever it says about positioning into payrolls is positioning as of seventy-two hours before payrolls. We will mark it on Monday.
What this does not tell you
The volatility and risk-reversal figures are single-sourced to one publisher’s option note and we did not find a second reader for them. If that page is wrong, the whole first half of this article is wrong. The note gives no spot reference of its own and no notional, so we cannot tell you how much was traded to move the front end 0.30 of a point — it could be a modest clip in a thin book.
We have no option-expiry ladder for today’s 14:00 UTC New York cut. The page this desk habitually reads does not resolve for today’s date, and the publisher that prints notionals puts its ladder out at about 11:40 UTC, ten minutes after this slot begins. That is now the twelfth consecutive slot with no strike-and-size data, and the reason is structural rather than bad luck: no daytime slot on this schedule can read a same-day ladder.
The 91-pip figure is a lognormal one-standard-deviation approximation from a single implied volatility quote. It is not a prediction, it is not a range the day will respect, and it says nothing about the path — a 91-pip day can be one move or four. The consensus figures are five different numbers from five publishers and we have not adjudicated between them, deliberately. And the October Federal Reserve pricing we are working from is 25 per cent on CME’s FedWatch as read at 03:35 UTC, which is a feed, at a timestamp, and will have moved by the time you read this.
Related
- FX event calendar — the release windows in UTC
- Signals — how we frame exposure around scheduled risk
- EA presets — filter and sizing configurations
