What is Antevo watching at the long end of the curve?
A spread that widens while the long end leads says the term premium is being charged rather than held down, which is the half of this position's falsifier that can be read off a screen. The half that cannot — what a sale actually clears at — is still unanswered: the 7 and 8 October auctions were named as the test and their result is not yet in the record.
**Marked to market, 8 October — the spread leg of the falsifier has fired; the auction leg cannot be read.** The falsifier here was an auction clearing above the screen with the thirty-to-five spread widening. The spread half has happened, and decisively: 47 basis points on 18 September narrowed to 40 on 23 September and then widened to a little under 58 on 2 October, the thirty-year at 5.630% against a five-year at 5.055%, with the long bond reaching 5.637% on 30 September — called then its highest since June 2002 — and 5.660% by 5 October. The compression this position tracked from 81 to 57 to 47, and on to 40, has reversed. The auction half is the problem. The 25 and 26 September editions named the ten-year and thirty-year sales of 7 and 8 October as the test, and no edition through 8 October carries their result. So this is published as it stands: the screen has answered and the saleroom has not, and the distinction this position exists to draw is precisely the one still missing.
Read the rest
**Marked to market, 20 September — the spread compressed a further ten basis points.** The thirty-to-five spread stands at 47 basis points, from 57 on 12 September and 81 in late August. The thirty-year settled 5.33% and the five-year 4.86%, the latter back at its decision-day close. The stated falsifier — an auction clearing above the screen with the spread widening — has not fired, and the spread has moved further from it in the interval. What has changed is the reason the auction matters. July net long-term capital flows printed negative against a large positive consensus, with foreign investors as net sellers of United States long-term securities, which makes the clearing yield a measurement of who is left rather than of what the term premium is worth.
Marked to market on 12 September, and this one resolves. The position named the thirty-year auction clearing yield against the screen as the cleanest read on whether the term premium is demanded or merely quoted, and set the falsifier as a sale clearing THROUGH the screen with strong cover. The sale on 10 September cleared 5.308% while the thirty-year screen closed 5.36% — inside, not through. It was nonetheless 9.2 basis points cheaper than the 5.216% paid at the prior auction, so buyers were compensated more for the same paper even as they stopped short of the market level. The screen level this position tracked has itself moved from 5.19% to 5.36%, and the thirty-to-five spread has compressed from 81 basis points to 57 as the front end outran the long end.
The screen level has moved. The thirty-year closed 5.19% on 26 August, having printed a nineteen-year high of 5.311% on 17 August, and is one basis point lower across the week.
What makes the auction the test rather than the screen is that the week's move was not parallel. The five-year rose three basis points to 4.38% while the thirty-year fell one, compressing the thirty-to-five spread from 85 to 81 basis points. A long end that holds while the front cheapens is either being demanded or being sponsored, and a clearing yield is the only place that distinction is priced by somebody who has to take the paper.
The context has sharpened it. The week's most prominent institutional commentary concerned whether the Treasury should be purchasing its own long-dated debt at all, and the third major government bond market supplies a parallel: a Japanese thirty-year auction is scheduled for 3 September against a 3.937% prior, with Japanese credit unions already absorbing bond losses as domestic long rates rise.
The thirty-to-five spread narrowing back below 40 basis points with the thirty-year under 5.50% — that would say the September widening was the supply calendar rather than the buyer base, and hand the level back to the front end. The brief drew the same 40-basis-point line on 25 September.
The stop against the screen, the tail, and the dealer take-down share
Whenever the 7 and 8 October sales enter the recordFor your assistantThe same answer as structured fields — MCP, JSON-LD
{
"position": "thirty-year-auction-watch",
"question": "What is Antevo watching at the long end of the curve?",
"url": "https://antevo.ch/ask/thirty-year-auction-watch/",
"observed": {
"value": "5.630% against 5.055%",
"as_of": "2026-10-02",
"statement": "The thirty-to-five spread on 2 October, widened from 47 basis points on 18 September"
},
"reading": "A spread that widens while the long end leads says the term premium is being charged rather than held down, which is the half of this position's falsifier that can be read off a screen. The half that cannot — what a sale actually clears at — is still unanswered: the 7 and 8 October auctions were named as the test and their result is not yet in the record.",
"falsifier": "The thirty-to-five spread narrowing back below 40 basis points with the thirty-year under 5.50% — that would say the September widening was the supply calendar rather than the buyer base, and hand the level back to the front end. The brief drew the same 40-basis-point line on 25 September.",
"next_test": {
"when": "Whenever the 7 and 8 October sales enter the record",
"what": "The stop against the screen, the tail, and the dealer take-down share"
},
"revision": 5,
"published": "2026-10-08",
"editions": [
{
"date": "2026-10-08",
"title": "Fewer tankers than normal left the Gulf for a third day with data",
"url": "https://antevo.ch/executive/notes/2026-10-08/"
},
{
"date": "2026-10-06",
"title": "On Monday the Bovespa rose 7.7% to a record and the real gained 4.4%…",
"url": "https://antevo.ch/executive/notes/2026-10-06/"
},
{
"date": "2026-10-03",
"title": "The US added 29,000 jobs in September against about 90,000 expected…",
"url": "https://antevo.ch/executive/notes/2026-10-03/"
},
{
"date": "2026-10-01",
"title": "A milder US inflation number failed to lower long-term borrowing costs",
"url": "https://antevo.ch/executive/notes/2026-10-01/"
},
{
"date": "2026-09-26",
"title": "Iran offered a plan to reopen Hormuz within a week and Brent fell 2.14%",
"url": "https://antevo.ch/executive/notes/2026-09-26/"
},
{
"date": "2026-09-25",
"title": "The 30-year Treasury yield rose 6 basis points to 5.46%, its highest…",
"url": "https://antevo.ch/executive/notes/2026-09-25/"
},
{
"date": "2026-09-24",
"title": "The 5-year Treasury yield rose 16 basis points to 5.00% and the 10-year…",
"url": "https://antevo.ch/executive/notes/2026-09-24/"
},
{
"date": "2026-09-20",
"title": "A quarter-point went in and the ten-year came out at exactly five per…",
"url": "https://antevo.ch/executive/notes/2026-09-20/"
},
{
"date": "2026-09-12",
"title": "Oil above a hundred, a Fed hike four sessions away",
"url": "https://antevo.ch/executive/notes/2026-09-12/"
},
{
"date": "2026-08-27",
"title": "Crude gave back its war premium for a third session and every maturity…",
"url": "https://antevo.ch/executive/notes/2026-08-27/"
},
{
"date": "2026-08-14",
"title": "Thursday settled firm and settled narrow",
"url": "https://antevo.ch/executive/notes/2026-08-14/"
},
{
"date": "2026-07-30",
"title": "The Bond Market Called the Fed's Bluff",
"url": "https://antevo.ch/executive/notes/2026-07-30/"
}
],
"advice": false
}antevo.ch/ask/positions.json · antevo.ch/mcp