The 158.61 Break Lasted One Closed Session. Tokyo Handed Back 42 Pips.

A break that happened only where nothing was trading

Here is the sequence, and it is worth stating as a sequence rather than as a move. On 7 October this desk marked 158.51 as the level to watch. On Monday 12 October it went — at 09:54 UTC, in London, with Tokyo closed for Sports Day, the Treasury cash market closed for Columbus Day and Canada closed too. New York pushed it to 158.75. The pair closed, on one reader, back at 158.61.

Then the markets that were shut opened. The Tokyo session ran 158.65 to 158.19, and by 05:46 UTC the pair was in the 158.30s. A second publisher, reading a different feed, has it at 158.351 against a stated previous close of 158.61 — a 25.9-pip decline, which agrees with the first reader to within rounding.

So the whole of the break was produced inside a window in which two of the three cash markets that price this pair were closed, and it was given back inside the first window in which they were not. That is not a repricing. That is what a 68-pip range looks like when the people who would have taken the other side were on holiday.

Gone against one convention. Untouched against the other.

Yesterday this desk published that Monday’s break was convention-dependent, and that we had published it without saying which 200-day it settled. The give-back has exactly the same problem, in reverse, and we are not going to pretend otherwise.

Against the simple readings — 158.53, published 7 October and named as a simple moving average, and 158.54, published 10 October and named as nothing — the 158.19 low is 34 and 35 pips below the line. Clean failure.

Against the exponential reading this desk published on 8 October, just under 158.00, the 158.19 low is 19 pips above the line, and the pair has not closed below it since 1 October. Nothing happened. Eleven days of the same regime, and this morning is day twelve.

One indicator. One price. Two opposite statements about whether a regime changed, two days running, in opposite directions. If your filter references “the 200-day” without naming a convention, it has now given you a false break and a false failure inside thirty hours.

The yield print this test needed does not exist

The whole point of waiting for the cash market to reopen was to get a yield print that would say whether Monday was a repricing or a liquidity artefact. The print is not there, and the reason matters.

The Federal Reserve’s own constant-maturity series for the ten-year stops at 8 October, at 5.22 per cent, with 5.28 on 7 October and 5.27 on 6 October before it. There is no observation for 9, 12 or 13 October in the issuing series as we read it this morning.

Monday’s widely-repeated 5.24 per cent therefore sits two basis points above the last figure the issuer actually measured, and it was quoted on a day the cash market was shut. One vendor shows 5.20 per cent for today against a stated previous close of 5.24 and a stated daily change of minus 0.03 — when its own two figures differ by 0.04. That is a four-basis-point move reported as three, on a page whose starting value was never a measurement.

Add it up and the adjudicating variable in this test is one unmeasured holiday quote, one vendor snapshot and an arithmetic error. We are not scoring a break against that. We are saying the test cannot be run on the data available, which is a different and more useful answer than a wrong one.

And it is a dollar move with a yen bid on top

If you are tempted to read 158.19 as a yen story — Japanese data, intervention risk, the usual — look at the other two legs. Euro–dollar ran to 1.1177 in Tokyo and was near 1.1175 in the early European session, roughly 20 pips above Monday’s New York close of 1.1155. Euro–yen ran 176.98 to 176.70, with the low 24 pips below the 176.94 close one reader published.

So the dollar is lower against the euro and the yen is higher against both. The dollar leg is doing most of the work and the yen is adding to it. For sizing purposes that distinction is the whole thing: a broad dollar give-back is correlated across your book, and a yen bid is not.

The stated driver on both sides of the session is crude, and crude is the one number nobody will pin down. One Tokyo desk had West Texas recovering to the middle of 92 dollars at 05:08 UTC and the rebound “limited” thirty-eight minutes later. One European note had Brent around 105 dollars and West Texas 92 to 93 in one paragraph and oil “elevated near 106 dollars” in another, on the same page. We are not building a causal claim on that.

What London has to settle this with: an AI panel and a retired order book

The busiest liquidity hours of the day start at 07:00 UTC, and there is nothing scheduled in them that adjudicates a 34-pip failure of a moving average.

The one Federal Reserve appearance in the window is at 07:45 UTC, and the Board’s own October calendar describes it as a discussion on artificial intelligence at the Bloomberg New Economy Forum in New Delhi. Our file carried that appearance with the correct time and no topic, and had it marked as the slot where a December rate probability might be restated. On the Board’s own listing it is not a policy event at all. An event you hold as a time without a topic is an event you have not actually scheduled.

The Board’s calendar also settles something else we were carrying loose: the Fed chair’s week-ahead appearance is listed at 11:30 p.m. on 15 October, which is 03:30 UTC on Friday 16 October at a fireside chat with the IMF managing director in Bangkok — not the 12:30 UTC one reader gave us. We can state the conversion with confidence because two other entries on the same page, at 3:45 a.m. and 5:40 a.m., land exactly on the 07:45 and 09:40 UTC times we already held from other sources.

As for where the resting orders are, the one page that gave a London order-book picture this morning is from a publisher whose order-book characterisation this desk retired. The reason recurred: it places thick sell orders in the early 158s and describes them as sitting near the support line. Price traded into that band an hour later. We are not publishing those clusters as a measurement of anything, and we are naming the problem rather than quietly using the numbers.

What this does not tell you

It does not tell you the break was wrong. A level can fail on a thin session and be taken out properly a week later, and 158.75 is still the highest print this pair has made in October. What we are saying is narrower: the evidence that would distinguish those two cases is a yield print, and the yield print is missing.

The exponential 200-day reading we are leaning on is five days old. We published it on 8 October and we have not recomputed it this morning, so the claim that the pair has stayed above it since 1 October is as of that date, not as of now. This desk has still never computed a 200-day from a price series of its own; every value in this piece is a publisher’s.

The 158.19 low and the 158.65 high come from one publisher, read twice, thirty-eight minutes apart. The second publisher we have corroborates the current level and the previous close, not the extremes — and that second publisher carries its own spot price at two values, 158.351 in its quote box and 158.3370 in its own prose, 1.4 pips apart. Treat the extremes as one house’s reading.

And the Federal Reserve calendar entries carry no timezone on the page itself. We have converted them as Eastern Time because two of them reconcile exactly with times we hold from independent sources, which is good evidence and not a published statement.

Related

• Economic calendar and event windows — the 07:45 and 12:30 UTC windows for the rest of this week.
• Signals — range-versus-trend character on the yen crosses.
• EA presets — filter settings for levels whose definition is a publisher’s choice.