How DeepStack settles a question · Part 07

Why did the long end climb?

Why did the long end climb after the hike: growth, debt, or a missing buyer?

From Part 07: The Fed Hiked. The Long End Still Climbed.. 10 voices on 3 sides, attributed and dated; the DeepStack Call with its history; 4 dated tests that settle it; the results as they come in.

The mechanism

The Fed raised short rates on September 16 and the 10-year Treasury still yielded 5.18 percent six trading days later, because the long end is priced by supply rather than by the policy rate. The federal deficit ran $1.97 trillion through August 2026 and the hyperscaler backlog stands near $2.3 trillion, so extraordinary capital spending and federal borrowing lean on the same bond market, while the Fed's short rate reaches only the households and borrowers it touches directly. Growth explains why the move is orderly; it does not make the refinancing bill disappear, and CleanSpark's 8.25 percent coupon shows today's rates reaching AI's borrowers. The Call would change if long yields fell while growth, inflation and AI capital spending stayed firm, or if Japanese and other foreign buyers returned in size without a better hedge.

Approved by Marco Cauduro on October 10, 2026.

The DeepStack callSkeptical

The bond market is pricing the AI bill, not a recession. The Fed cannot hike its way to a lower 10-year.

Supply sets the long end · Conviction High (4/5)

Our Call takes the side “A missing buyer”.

Read the full call

Call history2 versions · last changed
  1. v2Supply sets the long endConviction High (4/5)Side: A missing buyerSide named by the owner.
  2. v1Supply sets the long endConviction High (4/5)First call, as published with the story. The conviction, the actions and the flips were set down for the site on October 7, 2026.

6 voices

Growth: yields are catching up

  1. Michael Howell

    The current bond market cycle is behaving like a normal cycle, and a major buying opportunity for fixed income is approaching as yields near a peak.

    Source

  2. Michael Howell

    The risk-return balance for bonds is improving: a further 50bp yield rise would still give bonds a positive annual return, while a subsequent 100bp yield fall would generate a more than 12% annual return.

    Source

  3. Matthew C. Klein

    Contrary to what many officials expected, inflation stopped decelerating in 2023, which is especially visible in underlying measures that exclude volatile and idiosyncratic components.

    Source

  4. Matthew C. Klein

    The 'supercore' version of the PCE price index was rising about 1 percentage point faster than in the years immediately preceding the pandemic even before the surprises of 2025-2026, and is now rising about 2 percentage points faster.

    Source

  5. Kris

    Interest rate hikes are not currently a problem for the economy or investing, because if the economy and company profits grow faster than the interest rate, value is created, and the estimated Q3 GDP growth of 5.1% plus inflation gives nominal…

    Source

  6. Kris

    The main reason for higher inflation is probably the more expensive oil, and a higher interest rate will not cure that; the Fed just wants to ensure inflation doesn't get out of hand.

    Source

9 voices

Debt: 99% of GDP, up from 35%

  1. TSCS

    The Federal Reserve faces a trap: if it hikes rates it loses control of the long end because it feeds an annual interest bill approaching $1 trillion, but if it cuts despite 3.4% inflation it also loses the long end, creating a frustrating setup…

    Source

  2. TSCS

    TSCS argues governments and central banks do not want long yields higher and will go a long way to stop it, which means that if they succeed real yields fall and today's 2.65% over inflation won't be on offer for long, and if the way they succeed…

    Source

  3. Vitaliy Katsenelson

    Bond investors are rejecting the government-set price for Treasuries because of large, unaddressed budget deficits that may grow rather than shrink over the next 10 years, and because budget deficits imply inflation; Vitaliy Katsenelson says they…

    Source

  4. Vitaliy Katsenelson

    In mid-August 2026 the 30-year Treasury yield reached its highest level in 19 years, above 5.3% (last seen in 2007), and the 10-year yield hit 4.75%; these are important because 30-year mortgage rates are set from them and about two-thirds of US…

    Source

  5. Steve Eisman

    Treasury Secretary Bessent's $4 billion to $6 billion effort to buy long Treasuries and push rates down failed, as rates went up anyway, and his speaking at a partisan political convention undermined his credibility with the bond market.

  6. Steve Eisman

    Ten-year Treasury yields have broken through 4.8% and then 4.9%, the highest level since November 2023, driven by the Iran war pushing oil prices above $100.

  7. George Noble

    No secretary talks down $40T of debt.

  8. Stephen Clapham

    Stephen Clapham states they would maintain a near-zero weight in US Treasuries, other than those trading at a near-50% discount to nominal value which they see as an asymmetric bet, a position they say they have covered in recent weeks.

    Source

  9. Stephen Clapham

    A rise in short rates would be a problem given the government funds around 85% of its financing needs (the flow) with short-term bills, though on a stock basis bonds of over 5 years duration are only about 30% of the total, still a fairly…

6 voices

A missing buyer

  1. Michael W. Green

    The structural buyer base for long-duration Treasury debt has been withdrawing in stages for years—the Japanese hedged bid was damaged by the 2022 rate shock, Russia and China ceased buying in summer 2022, corporate pensions largely finished their…

    Source

  2. Michael W. Green

    What markets are showing is not a shortage of safe assets but a shortage of buyers willing to own duration, with sovereigns and corporates both saying the same thing: the corporate safe-asset factory is becoming ordinary risky debt and the…

    Source

  3. Mia Silverio

    Mia Silverio argues that persistent inflation concerns have pushed the U.S. 10-year Treasury yield near the closely watched 5% level for only the second time since the 2007-2008 financial crisis, and notes long-term Treasurys have returned -2%…

  4. Mia Silverio

    Canada's public pension funds are the third-largest pool of retirement savings in the world, and Canadian institutional managers are cooling on US investments: this past July Canadians sold $22 billion in US stocks, the biggest monthly sell-off…

    Source

  5. TSCS

    Oil explains the day-to-day wiggles in yields but not the trend, since Brent is cheaper today than at its spring peak while the 10-year is 70bp higher, even though since the war a $10 Brent move has come with about 5bp on the 10-year (roughly 3…

    Source

  6. TSCS

    TSCS argues the front end cannot cap the long end: a hike is only about 4/10 of a percent of receipts and mostly lands as income with people who spend it, Treasury buybacks manage volatility rather than level, and the buyers who used to close the…

    Source

2 voices

Voices that cross sides

  1. TSCS

    Kevin Warsh is a pretend hawk rather than a constrained one, meaning he talks tough on inflation while avoiding costly action, and this position can be falsified by a rate hike on 16 September.

    Source

  2. TSCS

    TSCS's forward view is that oil and long yields keep rising until stocks fall, because the S&P is the only release valve left: the bond-market valve that worked in April 2025 is the one the buybacks are designed to keep shut.

    Source

Where they cross

Who answers whom.

The specific points where one voice meets another: the same evidence read two ways, or the same mechanism with a different sign.

Alternative explanations

Four other readings.

What would have to be true for the Call to be wrong, as the other side argues it.

  1. No buyer has gone missing: US term premia are broadly stable and recently lower, with no sign of rising against foreign equivalents, so the loss-of-confidence reading fails (the case Michael Howell makes).
  2. The long end is pricing inflation, not the AI bill: inflation stopped decelerating in 2023, and supercore PCE now rises about two points faster than before the pandemic (the case Matthew C. Klein makes).
  3. Yields are catching up with growth: with third-quarter GDP growth estimated at 5.1 percent plus inflation, a higher rate is no problem for an economy and for profits that grow faster than it (the case Kris makes).
  4. It is the debt, not the buyer: a hike feeds an annual interest bill approaching $1 trillion and a cut at 3.4 percent inflation loses the long end too, so the Fed is trapped either way (the case TSCS makes).

What settles it

The dated tests.

The tests the story set, each with its specification where the owner has written one and the piece of the argument it verifies; on the Docket, resolved on the Results page.

  1. The Fed meets; another hike would support Klein’s inflation reading, while no move would strengthen Paulson’s case that September was a mistake.

    Part 07: The Fed Hiked. The Long End Still Climbed. Pendingp07-t01Pre-specifiedChecks the inflation claim

    Question
    Does the Fed funds target, upper bound (FRED series DFEDTARU) read higher on its first observation after October 28, 2026 than on its last observation before it?
    Source
    FRED, Fed funds target, upper bound (DFEDTARU)
    Rule
    The answer is yes if the reading after the date is higher than the reading before it (a hike); no if it is equal or lower.
    If the source is late, revised or silent
    FRED dates a change at its effective date, so the first observation after the date is the reading; a reading not in yet waits for the next run. A series that is late, revised or silent is inconclusive, and the Result says what was read and what was missing.

    The Call, before the dateEither answer fits the Call. Gives the yes answer 40%.

    40%

    0 · no100 · yes

  2. Treasury publishes August foreign holdings; renewed Japanese buying despite poor hedged returns would undercut the missing-buyer thesis behind the long-end selloff.

    Part 07: The Fed Hiked. The Long End Still Climbed. Call heldAnswer: nop07-t02Checks the missing-buyer claim

    Question
    Did Japan's holdings of U.S. Treasury securities at the end of August 2026, in the Treasury International Capital table of major foreign holders published in mid-October, read higher than at the end of July 2026?
    Source
    U.S. Treasury, TIC Table 5: Major Foreign Holders of Treasury Securities (monthly text table)
    Rule
    Compare Japan's August 2026 line with its July 2026 line in the same table. The answer is yes if the August figure is higher; no if it is equal or lower.
    If the source is late, revised or silent
    A table published after November 30, 2026, or a July figure revised in the same table so that the comparison changes sign, is inconclusive, and the Result says what was read and what was missing.
  3. 10-year at 5.5%

    Howell expects a major bond opportunity; failure to turn would keep the growth brake unproven.

    Part 07: The Fed Hiked. The Long End Still Climbed. Pendingp07-t03Open questionChecks the growth-brake claim

    Specification pending

  4. Leveraged-loan and corporate maturities arrive; stress would show today’s rates are reaching AI’s borrowers.

    Part 07: The Fed Hiked. The Long End Still Climbed. Pendingp07-t04Open questionChecks the refinancing-bill claim

    Specification pending

The test came in

Call held: the result is the one the Call expected.

Resolved October 19, 2026.

Every result is listed on the Results page.

Call heldAnswer: no

· Part 07: The Fed Hiked. The Long End Still Climbed. · Test p07-t02

Japan's holdings of U.S. Treasury securities fell: $1,103.9B at end-July 2026, $1,068.3B at end-August 2026.

The test. Treasury publishes August foreign holdings; renewed Japanese buying despite poor hedged returns would undercut the missing-buyer thesis behind the long-end selloff.

Treasury's Table 5 of major foreign holders, read on 2026-10-19, puts Japan's holdings at $1,068.3B at end-August 2026 against $1,103.9B at end-July 2026, a change of -$35.6B in the same table (U.S. Treasury, TIC Table 5: Major Foreign Holders of Treasury Securities).

The rule compares the 2026-08 column with the 2026-07 column for Japan: the later figure is not higher, so the test resolves refuted.

Japan, 2026-07 column (billions of dollars)
$1,103.9B
U.S. Treasury, TIC Table 5: Major Foreign Holders of Treasury Securities · Oct 19, 2026
Japan, 2026-08 column (billions of dollars)
$1,068.3B
U.S. Treasury, TIC Table 5: Major Foreign Holders of Treasury Securities · Oct 19, 2026

The DeepStack call. Conviction unchanged; the Call stands as written.

The result in full: what we said, the rule, what happened, what changes

What changed

The record of Part 07.

The Part’s publication, every change to the Call, every result that came in and the next dated test, oldest first.

  1. PublishedPart 07 published; the Call: Supply sets the long end.
  2. Call updatedThe Call: Supply sets the long end. The bond market is pricing the AI bill, not a recession. The Fed cannot hike its way to a lower 10-year. Our Call takes the side “A missing buyer”. by Marco Cauduro
  3. ResultCall held: the result is the one the Call expected. Call heldAnswer: no by Marco Cauduro
  4. Next testNext: Oct 28, 2026, The Fed meets; another hike would support Klein’s inflation reading, while no move would strengthen Paulson’s case that September was a mistake. Pre-specified
Sources (8)
  1. Federal Reserve - September 16 policy statement
  2. Federal Reserve - September 16 press-conference transcript
  3. FRED - 10-year Treasury constant-maturity series
  4. The Overshoot - rising bond yields analysis
  5. Potential Multibaggers - growth and the Fed hike
  6. JunkBondInvestor - who the hike hits
  7. HyperTech Investor - AI rally math
  8. Variant Perception - inflation and energy

Market data are as of the dates cited. Voices are paraphrased from public writing and attributed by name.

Analysis and opinion, not investment advice. Voices are paraphrased from public writing and attributed by name; DeepStack shows no score, ranking or accuracy for any person. Every question, one page each.