Four coupon securities settle today. A two-year note auctioned on 26 October, a five-year note on the 27th, a two-year floating rate note on the 28th and a seven-year note on the 29th, all announced together on 22 October, all landing on the Treasury’s books this morning. No bills settle today at all. The only document that shows those four rows side by side is the issuer’s own Tentative Auction Schedule, and it carries no offering amount, no auction time, no revision date, and nowhere in it does the word refunding appear — including on the three rows everybody will call the refunding on Wednesday.
What the issuer actually publishes, and what it leaves to you
The Tentative Auction Schedule of U.S. Treasury Securities has four columns: security type, announcement date, auction date, settlement date. That is the whole document. For today it gives the 2-Year NOTE, the 5-Year NOTE, the 2-Year FRN and the 7-Year NOTE, each announced Thursday 22 October 2026, auctioned Monday the 26th, Tuesday the 27th, Wednesday the 28th and Thursday the 29th respectively, each settling Monday 2 November 2026. The two-year floating rate note row is not marked R, so it is not flagged as a reopening.
What it does not give is any number you could size a position with. No amounts. No times. And no printed as-of or revision date anywhere on the page, which means a cached copy and a current copy are indistinguishable from each other. We have spent six weeks on this channel cataloguing calendar fields that carry a false precision; this is the other failure mode, and it is the better one. A blank field tells you to go and find the number. A field containing last quarter’s number does not.
Today also auctions the 13-week and 26-week bills, announced 29 October, settling Thursday 5 November. So the paper that settles this morning and the paper that is sold this afternoon are different objects on the same calendar day, and the schedule gives you no hour for either.
We can price two of the four, and our own published total is two billion light
A second reader gives the two-year note auctioned 26 October at $69B on offer and the two-year floating rate note auctioned 28 October at $30B. For the five-year and the seven-year we are carrying $70bn and $44bn from our own earlier articles rather than from anything read this morning.
Add those four and you get $213bn. This desk has published $211bn as the size of today’s settlement stack. The two-billion gap sits entirely in the floating rate note: our carried total implies $28bn where the reader we can read today says $30bn. So one of two things is true — either the FRN figure behind our $211bn was two billion low, or the reader is. We cannot settle it, because the one document that would settle it is the Treasury’s own offering announcement for 22 October, and that address is blocked to us by the issuer’s robots file.
Which means we are not asserting $213bn either. Two of its four terms are our own unre-verified carry, and a total built half from a source and half from your own file is not a measurement. The honest statement is that today’s coupon settlement is somewhere a little over two hundred billion dollars, that two of the four amounts are second-sourced this morning and two are not, and that the number we printed before was arrived at with at least one term we can no longer reproduce.
Wednesday is the quarterly refunding and the schedule will not say the word
On Wednesday 4 November the schedule shows three announcements: a 3-Year NOTE auctioning Monday the 9th, a 10-Year NOTE marked R auctioning Tuesday the 10th, and a 30-Year BOND auctioning Thursday the 12th, all three settling Monday 16 November. Three tenors, one announcement date, the classic shape. The document never labels them a refunding, never uses the word anywhere, and carries no note on that date.
That matters for anything automated. If your event schedule has a row called “quarterly refunding announcement” keyed off the issuer’s schedule, the issuer does not publish that row. You are either inferring it from the pattern of three coupon announcements on one Wednesday — which is what we are doing here, and we are saying so — or you are taking it from a vendor who inferred it for you and did not tell you they had.
The only other label the document carries for the next fortnight is a holiday: Wednesday 11 November, Veterans Day. That sits between Tuesday the 10th, when the 17-week, 4-week and 8-week bills settle, and Thursday the 12th, when the 6-week, 13-week and 26-week bills do. A settlement calendar with a hole in the middle of it is worth looking at before you assume next week’s cash flows are evenly spread.
The auction hour moved this weekend, and the schedule has no hour to move
Our own Systems desk published seven hours ago that it had not re-verified the hour at which Treasury auctions now announce and settle, and was explicitly not asserting it. We can close most of that now, by arithmetic off observed timestamps rather than off an issuer page.
The result page for the 26 October two-year note is stamped 26 Oct 2026, 17:03 UTC. The result page for the 28 October floating rate note is stamped 28 Oct 2026, 15:35 UTC. Those are consistent with a 13:00 Eastern coupon auction and an 11:30 Eastern bill-convention auction under Eastern Daylight Time, which was UTC minus four. From this morning Eastern Standard Time is UTC minus five, so the same two conventions are 18:00 and 16:30 UTC. Today’s 13-week and 26-week bills, on that reasoning, are a 16:30 UTC event and not a 15:30 one.
Be clear about the quality of that close. It is our subtraction from two publisher timestamps, not a published auction time, and neither of those pages labelled its stamp as the auction hour. We are closing the item to one decimal place of confidence, not to the minute. What we are not closing is the settlement hour, because nothing we read today gave one.
One number this week did cross-check cleanly
The 26 October two-year came in at a 4.708 per cent high yield with a 0.3 basis point tail, a 2.59 bid-to-cover against a 2.64 average, and allotments of 45.7 per cent indirect, 41.3 per cent direct and 13.0 per cent dealer. This desk filed the tail, the indirect share, the dealer share and the cover from a different publisher last week, and four of those five fields match to the decimal across two readers.
We are flagging that because it is rare enough on this channel to be information. In a week where the issuer publishes no amounts, the offering announcement is robots-blocked and our own carried total does not reconcile, the auction result itself is the one object that two independent publishers describe identically. When you are deciding which fields in your own store to trust, trust the ones that have survived a two-publisher comparison, and treat the rest as provisional however long they have been sitting there.
Why this is a sizing question and not a view
Lay the week out and the shape is front-loaded on paper and back-loaded on data. Just over two hundred billion dollars of coupons settles this morning. Three more tenors are announced on Wednesday. The ISM manufacturing survey prints tonight at 15:00 UTC, an hour later than its October self. The Reserve Bank of Australia decides at 03:30 tomorrow, the same day Americans vote. The Bank of England is Thursday at 12:00, and payrolls is Friday at 13:30.
So the supply hits the balance sheet before the week’s two largest data points tell anyone what it is worth. That is not a directional argument and we are not making one. It is an argument about when your yield-sensitive exposure is carrying risk it is not being paid for: today and Wednesday are mechanical events with no information in them, and Thursday and Friday are information events with no mechanics. A book sized evenly across the five days is overweight the first half.
What this does not tell you
We read no offering amount at the issuer. Not one of the four amounts above comes from a Treasury page: two come from a secondary reader this morning and two from our own file. The Treasury’s own 22 October offering announcement is the document that would settle all four, and it is refused to us by robots.txt, so every dollar figure in this article is a secondary or carried reading of an issuer document we could not open.
We did not confirm the 2 November settlement date at any second source — it comes from the Tentative Auction Schedule alone, which is the issuer and is therefore good, but it is one reading. The auction hour is our arithmetic. The refunding label on Wednesday’s three rows is our inference from their shape. We have no when-issued level for anything this week and no dealer-positioning data at all, so nothing here says whether today’s settlement is being absorbed easily or not. And we did not read a single price in this article: there is no yield level, no spot level and no curve shape in it, deliberately, because the schedule does not contain any.
