Dollar–yen traded 157.987 to 158.764 through the New York session, and the high is the strongest level since 14 October. Somewhere inside that range it went through the 200-day moving average, which one Tokyo publisher puts at about 158.60 and says has capped the pair since late September. The break happened in the hours before the first American polling station closed. That ordering is the whole story: the structural level that mattered gave way without the catalyst that was supposed to move it, and the modal outcome of that catalyst — as the same desks writing this morning describe it — points the other way.
The level went before the event, and nobody will name what took it
The 200-day line is a single-sourced number. One Tokyo wire, writing at 19:55 UTC, put it near 158.60 and said the pair was testing a break above it, with price at 158.71 at that moment. We have not found a second publisher giving a level for that average, so treat 158.60 as one desk’s arithmetic rather than a fact, and treat the distance from it — 16 pips at the session high — as correspondingly soft.
What is not soft is the direction of the explanation. The same wire says plainly that there was no dollar strength in this move, that yen weakness pushed the pair higher, and that there was no direct catalyst for the yen weakness. This desk has a standing way of handling that: the move is real, we looked for the driver, we did not find one, and we say all three rather than inventing a cause to fit the candle.
The candidate drivers are all available and none of them survives contact. The session low at 157.987 came, on one reading, after reports of progress between Washington and Tehran on nuclear and frozen-asset questions, with crude falling and long yields dropping alongside. But the same day carried a reported American strike on Qeshm Island, sourced to unnamed officials, and an Iranian official quoted as saying the two sides were far from a deal. Brent was $101.54 at 14:00 UTC and down about three per cent, near $104 by late morning, and near $103 at the open and down one per cent — three levels, one session, and Monday had already taken it down more than seven per cent. If crude was the mechanism it was a mechanism pointing in both directions within the same eight hours.
The yield rise that supposedly did it is not a number we have
The cleanest version of the bullish case is that the September job-openings print at 15:00 UTC lifted American yields and dragged the pair with them. One Tokyo headline says exactly that. We will not publish the size of it, because we cannot read it anywhere.
Two readers put the ten-year near 4.91 per cent at the New York open. One briefing carries 4.909 per cent as the previous session’s close and a live board on the same page showing 4.9950, which is a 9-basis-point internal disagreement on one publisher’s page and a reason to use neither. Nobody we could open publishes where the ten-year actually settled on 3 November. So the proposition on offer is that the yen fell because yields rose, from a level that was roughly where it had been the day before, by an amount no reader states.
That is not a reason to reject the explanation. It is a reason to refuse to size off it. If your rule is “add risk when the ten-year repriced more than N basis points,” you cannot evaluate the condition tonight, and a rule you cannot evaluate should not fire.
The count started at the Tokyo open, and three readers describe the Senate three different ways
Counting began at 23:00 UTC, which is 08:00 in Tokyo — the minute the session this article is written into opened. Yesterday we published that the first polling stations close at 23:00 UTC and the last at 06:00 UTC on eleven closing boundaries agreed by two independent readers, and that the only dated resolution step on any calendar for one of tonight’s races is an Alaska tabulation on 18 November. Nothing this morning changes that. One Tokyo desk says outright that a clear picture could take several days, because close races need mail ballots postmarked by election day.
What the three morning readers do not agree on is the race itself. One calls the Senate close. One says that if polls hold the Democrats take it 53 seats to 47. A third cites prediction-market traders at roughly a 75 per cent chance of a Democratic Senate. Those are three characterisations of one binary, and we published the argument against all of them yesterday morning: a probability on a binary outcome is silent about the margin. The Reserve Bank of Australia made the point for us, hiking exactly as four feeds said it would, with three of its nine members voting against.
Here is the part that should bother a trend-follower. The base case the same desks publish is a Democratic House and a divided Congress, and they call that dollar-negative and yield-negative. The surprise case — Republicans holding both chambers — is the one that takes the pair at 159. A Democratic sweep puts it in the low 157s. So the price has just broken up through its 200-day line, and the consensus scenario for the event it broke ahead of would send it back down through the level it just cleared. One desk’s forecast range for today is 157.700 to 159.300: 160 pips wide, straddling the break, which is an honest way of saying the level carries no information tonight.
The crowded side has not moved, and the measurement is eight days old
One morning note says the IMM yen futures position shows a net yen long above 100,000 contracts. Yesterday morning this desk published 101,591 contracts as of 27 October and flagged that it was a Tuesday-close measurement already six days old. It is now eight days old, and today’s reading is a second publisher restating the same level without refreshing the date. That is confirmation of the level and no information at all about the week since.
So the standing position into a multi-day count is a net long in a currency that has just lost its 200-day line, measured before any of this happened. One desk expects dips bought in the upper 158s, is cautious above 159, and says intervention concern rises near 160 — all single-sourced, all one house’s view, and the 160 figure in particular is a round number somebody chose rather than a level anybody has defended.
The practical consequence is about the fixing, not the direction. Tokyo was shut for Culture Day yesterday, there was no 09:55 fixing, and the session traded 18 pips. Today the market is open and the fixing is back at 00:55 UTC — inside the count, with results arriving in waves behind it. A fixing-window rule calibrated on ordinary days is being asked to run through a session whose event risk does not resolve at a known minute.
What this does not tell you
We cannot give you yesterday’s close. One desk reports the pair finished 0.5 per cent higher on a close basis with a peak around 158.76; another has it near 158.75 late with a 158.71 snapshot; a third’s headline puts it in the upper 158s and up 75 sen. The page that would settle it — the four-values table for 3 November — would not load for us, and neither would the one broker report whose headline carries the 75-sen figure. So we name the band, 158.71 to 158.76, and refuse the single number. A gap-fade rule sized off whichever of those arrived last is sized off an accident.
We also have no index levels for the 3 November American close. The one page that covers the session gives exchange-traded fund moves at the bell — the Nasdaq-100 tracker up 0.62 per cent, the S&P 500 tracker up 0.31, the Dow tracker down 0.44, with Apple down 1.90 after a gross-margin forecast — and no index closes at all. That is enough to know the tape was mixed and not enough to put a number on it.
The 200-day level, the forecast range, the intervention threshold and the futures characterisation are each one publisher’s figure. The crude levels are intraday prints at stated times, not settlements. And nothing here tells you which way the count goes, because at the hour of writing almost nothing has been counted.
Related
- FX event calendar — the 23:00 UTC start, the 00:55 fixing, and tonight’s American cluster
- Signals — how a level break with no sourceable cause is handled
- EA presets — news-window and fixing-window settings for a session with no known resolution time
