INVESTMENT NOTE NO.14OCT 21, 2026FOMC Preview · U.S. Treasuries

The U.S. 10-Year Yield Tends to Peak
Near the Final Rate Hike

At the October FOMC, the question is less whether the Fed holds than how much tightening remains—and for how long.

In five of the six Fed tightening cycles since 1990, the 10-year Treasury yield reached its peak within four months of the final rate hike. But the final hike can only be identified with hindsight. This time, Treasury supply and demand for capital to finance AI investment are also shaping long-term yields.

U.S. 10-year · Oct. 20: 5.12%+94bp YTD (U.S. Treasury; Snowshagal calculations)¹
Fed funds target range: 3.75–4.00%+25bp on Sept. 16; first hike since 2023²
September dot-plot median: 4.1%Same at year-end 2026 and 2027 · 4.00–4.25% target range³
FOMC decision: Oct. 28 · 2pm ETOct. 29, 3am KST · No new SEP⁴

FIGURE 1

How close to the final hike did 10-year yields peak?

▲ Six Fed tightening cycles since 1990. Each peak is the highest daily 10-year Treasury constant-maturity yield from the first hike through two years after the final hike (Fed H.15; a one-year post-hike window produces the same peaks). The 2004–06 cycle's 2007 peak was only 1bp above the 5.25% reading three days before the final hike (hollow marker). Month offsets are calculated from FOMC decision dates. An observation, not a forecasting rule.^{5, 6, 7, 8, 9}

AT A GLANCE · FIVE TAKEAWAYS

The historical pattern is five out of six. Across six post-1990 tightening cycles, the 10-year yield peaked one to four months before the final hike in 1994, 2000 and 2018; 17 days after it in 1997; and roughly three months afterward in 2023 (Snowshagal calculations). Only the 2004–06 cycle peaked almost a year later—and that reading exceeded the yield three days before the final hike by just 1bp. This is a historical observation, not a prediction.
The number of hikes is not the answer. Cumulative tightening by the time of the yield peak ranged from 25bp in 1997 to 525bp in 2023 (Snowshagal calculations). The Fed hiked 25bp on September 16. Whether this is a brief recalibration or the start of a longer tightening campaign remains unresolved.
The dot plot does not say December will be the last hike. The September median is 4.1% at both year-end 2026 and year-end 2027. But the 2027 dots could reflect hikes early in the year followed by cuts later, as Governor Waller noted. Of the 18 participants, 16 projected at least one more hike in 2026, including four who projected two.
The Fed is not the only driver of long-term rates. In the Federal Reserve Board's Kim–Wright model, the term premium accounted for roughly half of the approximately 87bp rise in the fitted 10-year zero-coupon yield from end-2025 to October 16 (Snowshagal calculations). Treasury raised the size of its longer-dated liquidity-support buybacks to at least $4 billion per operation in September and is due to announce the next plan at its November 4 quarterly refunding.
There is no new dot plot on October 28. Watch the statement and press conference for clues to how much more the Fed might raise rates and how long it would keep them elevated. In particular, will Chair Warsh depart from the September minutes' assessment that financial conditions still support growth despite higher long-term yields? Read the market reaction against what was priced in before the meeting.

The U.S. 10-year Treasury yield stood at 5.12% on October 20, based on the Treasury's constant-maturity series—94bp above its level at the start of the year (Snowshagal calculations).¹ The Fed meets on October 27–28. The statement is due at 2 p.m. U.S. Eastern on October 28 (3 a.m. October 29 in Korea).⁴ To judge whether the 10-year yield is near a peak, what matters is less how many hikes have already occurred than how high—and for how long—the market expects rates to remain. The final hike is one clue to that path, but persistent term premia and Treasury supply pressures mean the end of a hiking cycle need not bring lower long-term yields. This note examines the historical evidence, the Fed's position today, signs of stress from elevated rates, and what to watch at the upcoming meeting.

Q1PEAKS · SIX TIGHTENING CYCLES

How close was the 10-year yield peak to the final Fed hike?

In five of six tightening cycles since 1990, the peak came within four months on either side of the final hike. What drove yields lower afterward varied: economic strain, credit and asset-market stress, and shifts in Treasury supply and demand.

Hana Securities strategist Heo Seong-woo summarized six major 10-year yield peaks since 1990 as reversals triggered by the “side effects” of high rates. Its summary said that, excluding 1990, the other five occurred within four months of the final rate hike.¹⁰ Those six peaks are a selection made by Hana Securities: they include a 1990 peak during an easing cycle and exclude 1997, when the Fed hiked just once. The same report identifies 2007 as an exception. On Hana's own selection, therefore, the count is four out of five tightening-cycle peaks.

This note takes a different approach. It includes all six Fed tightening cycles since 1990 and defines each yield peak as the highest 10-year rate observed between the first hike and two years after the final hike. The measure is the Fed's H.15 10-year constant-maturity Treasury yield—a daily observation derived from the yield curve, not the closing yield on a particular bond.^{5, 6} The two-year window captures peaks that formed while the Fed was on hold after its final hike and matches the horizon used for the table's “Two years later” column. Shortening the window to one year leaves all six peaks unchanged; shortening it to six months changes only the 2004–06 cycle's peak to 5.25% on June 26, 2006 (Snowshagal calculations). Five peaks fall within ±4 months under the two-year and one-year definitions; all six do under the six-month definition. In the exceptional 2004–06 cycle, the maximum was 5.26% on June 12, 2007. Yet the yield had already reached 5.25% three days before the Fed's final hike on June 29, 2006—a difference of just 1bp.⁷

Fed tightening cycles and 10-year Treasury yield peaks since 1990Peak = highest Fed H.15 10-year constant-maturity yield from the first hike through two years after the final hike; ties go to the first date reached (April 11/14, 1997; January 20/21, 2000). Final hike dates are FOMC decision dates. Month offsets, cumulative tightening, and two-year changes are Snowshagal calculations; two-year change = the yield at month-end in the same month two years later minus the peak. Events in the final column are Hana Securities' cited catalysts, reordered chronologically.
Cycle	10-year peak	Final hike	Peak minus final hike	Hikes to peak	Two years later	Backdrop around the peak · Events that deepened the decline
1994–95	8.05% · Nov. 7, 1994	Feb. 1, 1995	−2.8 months	9 months · 175bp	−199bp	Unwinding leveraged positions · A further 75bp hike on Nov. 15 after the peak; Orange County bankruptcy and the Mexican peso crisis in December
1997	6.98% · Apr. 11, 1997	Mar. 25, 1997	+0.6 months	17 days · 25bp	−162bp	One hike after 75bp of earlier cuts, then a pause · Asian financial crisis began in July, three months after the yield peak
1999–2000	6.79% · Jan. 20, 2000	May 16, 2000	−3.8 months	7 months · 75bp	−172bp	Treasury announced plans for up to $30bn in debt buybacks on Jan. 13, seven days before the peak, amid fiscal surpluses · Nasdaq peaked in March, followed by the dot-com bust
2004–06	5.26% · Jun. 12, 2007	Jun. 29, 2006	+11.4 months 5.25% three days before final hike	35 months · 425bp	−173bp	Peak almost a year into the Fed's pause · Bear Stearns hedge-fund distress around the peak; BNP Paribas froze fund redemptions in August
2015–18	3.24% · Nov. 8, 2018	Dec. 19, 2018	−1.3 months	35 months · 200bp + quantitative tightening	−240bp	U.S.–China trade war · December equity selloff and Powell's January 2019 pivot followed the peak
2022–23	4.98% · Oct. 19, 2023	Jul. 26, 2023	+2.8 months	19 months · 525bp	−87bp	Term-premium surge · Thirteen days after the peak, the Nov. 1 Treasury refunding slowed coupon-auction increases and Powell held a press conference; the December dots pointed toward cuts

In three of the six cycles—1994, 2000 and 2018—the 10-year yield peaked before the Fed's final hike. Hana Securities interprets this as the bond market turning when investors became confident that most of the tightening was already behind them.¹⁰

Putting Hana's proposed catalysts in chronological order shows that several occurred after the yield peak. In 2023, the 10-year peaked on October 19, while the Treasury's decision to slow increases in coupon issuance and Chair Powell's press conference came 13 days later, on November 1.¹¹ The 1994 Orange County bankruptcy and peso crisis, BNP Paribas's suspension of redemptions in 2007, and the December 2018 equity selloff likewise followed their respective yield peaks. Chronologically, they make more sense as forces that accelerated subsequent declines than as events that caused the peaks themselves. Nor does something have to break for long-term yields to fall: easing inflation or reduced expectations of further tightening can lower the expected path of short-term rates.

All six cycles saw the 10-year yield below its peak two years later, with declines ranging from 87bp for 2023 to 240bp for 2018 (Snowshagal calculations using month-end observations). But two years after the 2018 peak came the COVID-19 pandemic, while the period two years after the 2007 peak encompassed the financial crisis. These outcomes cannot be attributed to the initial yield peaks alone. Hana's average decline of 164bp uses a different set of dates and includes 1990.¹⁰

Q2CYCLE · WHERE THE FED STANDS

How far into a new tightening cycle is the Fed?

The Fed resumed hiking a month ago. That alone does not mean the 10-year yield peak is a long way off. We do not yet know whether this is a short recalibration or the beginning of a longer tightening cycle.

On September 16, the Fed raised its target range by 25bp to 3.75–4.00%—its first hike since 2023—on a unanimous 12–0 vote.² Hana Securities calculated that in five historical tightening cycles, the 10-year yield peaked an average of 21 months and a cumulative 280bp of hikes after the first increase.¹⁰ Those averages are not milestones this cycle must reach. In 1997, the 10-year yield hit 6.98% just 17 days after the Fed's single hike on March 25.^{5, 6}

FIGURE 2

Cumulative hikes before the peak: 25–525bp · In 1997, one hike was enough

Cumulative rate increases and elapsed time from the first hike to the 10-year yield peak

Cumulative tightening measures hikes implemented by the yield-peak date (the 1994 peak preceded the Nov. 15 75bp hike). The 2026 reading is as of October 20; whether the yield has peaked is unknown. Snowshagal calculations based on Fed decision dates and H.15 data.^{5, 6, 7, 8, 9}

SEPTEMBER FOMC MINUTES · RELEASED OCT. 7

Why did the Fed start hiking again?

The case for further tightening rests on these assessments, not on a predetermined number of hikes. Read subsequent housing and credit data for signs that those assessments may be changing.¹²

Inflation: Participants generally judged inflation risks to be tilted to the upside, with some seeing a greater tilt in recent months. They cited inflation pressure from geopolitically driven oil-price increases and a surge in AI-related investment.
Employment: The labor market was considered stable and close to maximum employment. More than half the participants viewed it as having strengthened somewhat recently.
Policy stance: Several participants viewed the current policy rate as not restrictive or only mildly restrictive. There were also references to upward revisions in estimates of the neutral rate.
Financial conditions: Many participants believed financial conditions were still supportive of growth despite higher long-term Treasury yields, citing strong equity gains, tight corporate credit spreads, and healthy loan and bond issuance. A few pointed to housing as an exception because of elevated mortgage rates.
Next steps: Most participants judged that one more hike before year-end would likely be appropriate, while stressing that each meeting should be assessed with an open mind.

2004–06 · GREENSPAN'S CONUNDRUM

Aggressive Fed hikes; little rise in long yields

+425bpFed funds rate

+50bp10-year yield (approx.)

A global savings glut, official purchases linked to exchange-rate management, and a term premium that fell close to zero kept long-term yields subdued.¹⁰

2026 · THE REVERSE CONUNDRUM

One small Fed hike; long yields have already surged

+25bpFed funds rate (Sept. 16)

+94bp10-year yield (YTD)

Competing demand for capital from AI infrastructure, defense and fiscal financing, more price-sensitive buyers, and a higher term premium are pushing yields up, according to Hana Securities.^{1, 10}

The comparison periods differ. The 2004–06 card covers the full tightening cycle (June 2004–June 2006); the 2026 card shows one Fed hike in September against the 10-year yield's change from the beginning of the year through October 20.

Hana Securities places the spread between the policy rate and the 10-year yield at about 110bp (up to 130bp), above its historical peak average of roughly 100bp.¹⁰ On October 20, the difference between the 10-year yield (5.12%) and the upper bound of the fed funds target range (4.00%) was 112bp (Snowshagal calculation).¹ That gap reflects both expectations of the future policy-rate path and the term premium. It cannot, on its own, tell us the contribution of either component.

A model-based decomposition offers a closer look. In the Federal Reserve Board's Kim–Wright estimates through October 16, the 10-year fitted zero-coupon yield rose about 86.7bp this year: 41.5bp came from the term premium and 45.2bp from the expected short-rate path (Snowshagal calculations). The model's 10-year term premium reached 1.10% on October 5, its highest since April 2010, before easing to 0.99% on October 16.¹³ Hana Securities described the premium as being at a 12-year high, but both the level and the historical ranking depend on the model used.¹⁰ The September minutes also attributed part of the rise in yields to a higher expected policy-rate path and strong economic data. They noted market commentary attributing the rise in term premia to geopolitical risks, uncertainty surrounding Treasury buybacks, and heavy corporate debt issuance to fund AI infrastructure.¹²

What the dot plot tells us—and what it does not

The September dot-plot median was 4.1% at both year-end 2026 and year-end 2027.³ The dots show year-end levels, not the path between them. Identical medians do not rule out hikes followed by cuts during 2027, nor do they identify whether the next hike would come in October or December.

Governor Christopher Waller addressed that distinction in an October 8 speech. Of the 18 participants who submitted September dots, 16 expected at least one more hike that year, including four who expected two. The 2027 dots are harder to interpret because they could combine early-year hikes with subsequent cuts, but Waller said eight participants projected rates 50bp above their then-current level at the end of 2027. He also emphasized that hikes need not come at consecutive meetings.¹⁴ A pause in October could therefore leave the terminal rate—and the length of time rates stay elevated—unchanged.

SEPTEMBER DOT PLOT · TARGET-RANGE MIDPOINT4.1%

Same median for year-end 2026 and 2027; equivalent to a 4.00–4.25% range. Does not reveal hikes or cuts during 2027.^{3, 14}

HANWHA INVESTMENT & SECURITIES · UPPER BOUND4.25%

One more hike in December, then hold. Consistent with the September 2026 year-end dot-plot median.¹⁵

HANA SECURITIES · UPPER BOUND4.50%

Two more hikes, in December and next March.¹⁰

U.S. 2-year · Oct. 204.79%

+132bp YTD (U.S. Treasury; Snowshagal calculations). At least a 90% market-implied probability of one or more additional hikes by the December meeting as of Oct. 19 (Kiwoom Securities).^{1, 16}

The 4.1% dot-plot figure is the rounded midpoint of the target range (4.125%), while the broker forecasts use the range's upper bound. Thus 4.1% in the dots and Hanwha's 4.25% refer to the same 4.00–4.25% target range.

The 2-year yield of 4.79% exceeds the brokers' projected terminal-rate upper bounds of 4.25–4.50%. That could mean markets are pricing more hikes, a longer period of elevated rates, or additional risk compensation in the 2-year Treasury yield. Based on market pricing he cited on October 8, Waller said the probability of at least two hikes by the March 2027 meeting was approaching 80%.¹⁴ Those numbers alone cannot distinguish among the explanations.

Expectations of an October hike have receded. In an October 12 report, Hana Securities said the probability had fallen from as high as 70% to roughly 20%, citing, among other factors, Waller's caution against assuming back-to-back hikes.¹⁷ Chair Warsh said at an IMF event on October 16 that he would not prejudge the meeting.¹⁸

Hana Securities views 1997 as the closest historical parallel. After cutting 75bp, the Fed hiked once and stopped; the 10-year yield peaked on April 11, 17 days later. Hana argues that the Asian financial crisis, which began in July of that year, strengthened the dollar and lowered import prices, easing inflation. But the crisis began three months after the yield peak: it is more plausibly a factor that deepened the subsequent decline than one that caused the peak. Hana sees no comparable external shock today.^{5, 10}

Q3WINDOW · CONDITIONAL COMPARISON

What if the final hike comes in December? Or next March?

This is a conditional comparison. If December is the final hike, the historical ±4-month window runs from August 2026 to April 2027 (Snowshagal calculations), encompassing the early-October 5.31% yield reading. If the last hike comes in March 2027, the window begins in November. Change the assumed policy path, and the window moves with it.

FIGURE 3

Transposing the historical ±4-month window onto 2026–27

This is not a probability-based forecast of when yields will peak. It simply places the historical observation window around two hypothetical dates for the final hike.

Shaded bands = four calendar months on either side of the assumed final-hike FOMC date (Snowshagal calculations). A follows Hanwha's December endpoint, consistent with the September 2026 median; B follows Hana's forecast. Meeting dates are from the Fed (2027 dates are tentative). Red marker = 2026 high of 5.31% on Oct. 5 (Fed H.15).^{4, 5, 10, 15}

The 10-year yield reached a year-to-date high of 5.31% on October 5 (Fed H.15), then fell 19bp to 5.12% by October 20 (same series; Snowshagal calculations).^{1, 5} In the Kim–Wright model, the fitted 10-year zero-coupon yield declined roughly 18bp from October 5 to October 16, of which about 11bp—or a little over 60%—came from a lower term premium (Snowshagal calculations).¹³ Under a scenario in which December marks the final hike, we are already within the historical observation window, leaving open the hypothesis that the 5.31% reading on October 5 was the cycle's peak (hereafter, the “early-October peak hypothesis”). Hana Securities expects the last hike in March 2027 but places the likely peak window around the December hike, broadly from Q4 2026 to Q1 2027, while flagging risks on both sides. A December dot plot signaling that rates will stay high well into 2027 could delay the peak; earlier signs of strain from high rates could bring it forward.¹⁰

If the timing of the final hike changes, so does the window. One of the six historical cycles, 2004–06, reached its maximum outside the window.

Q4STRAIN · THE COST OF HIGH RATES

Where is the strain showing—and is AI credit risk spreading?

The stress is evident in housing and parts of the AI sector. The broader market indicators reviewed here do not yet establish a credit crunch. But the stress reflected in today's spreads is a different question from the channels through which a future shock could propagate.

Hana Securities points to 2007, when the 10-year yield peaked a year after the final hike, as an example of stress emerging with a lag where leverage had accumulated—then, in housing finance. Potential trouble spots today include private credit, a surge in AI-related bond issuance, and commercial real estate.¹⁰ The following dashboard assesses information available through October 20 at three levels: individual firms and sectors, the broader market, and financing pressures.

High-Rate Stress Dashboard

Based on Oct. 13–20 data and reporting; observation dates vary by indicator

COMPANIES & SECTORS

CoreWeave CDSAI cloud · neocloud provider

Above 1,000bp

Yield on its April-issued bonds (9.75% coupon) exceeded 15% (Eugene Investment & Securities). Q2 2026 net interest expense was nearly 25% of revenue ($640m ÷ $2.575bn; Snowshagal calculation).^{19, 20, 21}

Under stress

Big Tech CDSAmazon · Alphabet · Meta · Nvidia

Near record highs

All four traded near record-high CDS levels on Oct. 15; individual quotes were not disclosed. Oracle's five-year CDS widened from 260bp to 288bp in one week (Reuters). Absolute levels were far below CoreWeave's; Eugene Investment & Securities' chart suggests roughly 100bp for some large issuers.^{19, 20}

Low but rising

NAHB Housing Market Index

October reading, lowest since June 2012. Mortgage rates have recently been well above 7%. September FOMC minutes also identified housing as an area where financial conditions were not supportive of activity.^{12, 22}

Acute rate pressure

Pending home sales · Housing startsSeptember · Reported Oct. 20

−7.3% · −5.2%

Pending Home Sales Index at 66.0, its lowest since the series began in 2001; down 11.3% year over year, with declines in all four regions (NAR). Housing starts ran at an annualized 1.221m units (Census Bureau; 90% confidence interval of ±11.8%, so the reported change was not statistically significant). The 30-year fixed mortgage rate was 7.40% on Oct. 15, the highest since November 2023 (Freddie Mac).^{23, 24, 25}

Housing weakness

BROADER MARKET

U.S. high-yield spread

331bpOctober high

Widened from around 260bp at end-August (ICE BofA: 263bp on Aug. 31) to 331bp on Oct. 16; eased to 329bp on Oct. 19 (Hana Securities; ICE BofA index). Still below the post-2019 median of 356bp cited by iM Securities. This argues against calling it a systemic credit crisis today, not against the possibility of contagion.^{26, 27, 28}

Wider, but below median

S&P 500From record closing high

−1.4%

7,818.93 on Oct. 6 versus 7,712.57 on Oct. 20 (Snowshagal calculation). Well short of the 8% drawdown from the peak that Hana Securities associates with periods when rising yields often stall or reverse.^{17, 29}

Below the cited threshold

FUNDING PRESSURES · COMPETITION WITH TREASURIES

AI-sector corporate bonds2026 issuance · 10-year Treasury equivalents

$187.1bn

Not the actual face amount issued: a duration-adjusted measure of interest-rate risk (issue size × modified duration ÷ 8, following the Dallas Fed methodology). Equivalent to approximately 52% of U.S. 10-year Treasury issuance over the same period (Hanwha Investment & Securities).¹⁵

Competing for capital

French 10-year yield

4.93%

Jumped more than 14bp on Oct. 19 (Kiwoom Securities). About 145bp above the German 10-year yield of 3.4776% (Snowshagal calculation).^{16, 30}

Global yield pressure

iM Securities economist Park Sang-hyun argues that, although CDS spreads have risen for CoreWeave and others, relatively stable U.S. credit spreads mean the risk of broader contagion is not yet evident.³⁰ Eugene Investment & Securities considers concerns about CoreWeave's credit quality warranted, but sees limited near-term bankruptcy risk in light of its remaining performance obligations (RPO) exceeding $100bn.¹⁹ That is the brokerage's credit-risk assessment, not proof that the backlog can service debt. CoreWeave's Form 10-Q shows $103.7bn of RPO at June 30: 41% expected to be recognized as revenue over the next 24 months, 39% over the following 24 months, and the balance in months 49–78. Borrowings on its balance sheet at the same date totaled approximately $35.1bn (Snowshagal calculation).²¹ A large RPO balance supports demand visibility, but it is not cash in the bank. The capital expenditure and debt required to fulfill those contracts must be considered alongside the backlog.

The IMF's October 13 Global Financial Stability Report examined 14 publicly disclosed AI-related transactions since 2023. Roughly $300bn of equity investment and $650bn of debt were linked to $1.4tn of future revenue commitments. If a central recipient of that financing falters, expected revenues and investment valuations could deteriorate together.³¹

Kiwoom Securities said upcoming Big Tech earnings, beginning with Meta on October 21 U.S. time, should be examined for signs that the sector can break a negative feedback loop: rising AI data-center investment, more outside borrowing, persistently high rates and wider credit spreads, higher financing and refinancing costs, and longer investment payback periods.³²

The indicators reviewed so far do not establish a broad market credit crunch. Pressure is concentrated in housing and selected AI companies, while supply-and-demand forces pushing long-term yields higher remain in place.

Q5COUNTERARGUMENT · WHY HISTORY MAY FAIL

Why this pattern may not hold this time

Three reasons warrant caution: the final hike cannot be identified in real time; roughly half of this year's rise in the fitted 10-year zero-coupon yield reflects the term premium in the Kim–Wright model (Snowshagal calculation); and the sample contains only six cycles.

REAL-TIME PROBLEM

You only know the final hike in hindsight

Most participants in the September minutes anticipated another hike by year-end, and Governor Waller noted that hikes need not come at consecutive meetings. In three of the six historical cases, the yield peak preceded the final hike; investors at the peak could not have known which hike would be last.^{12, 14}

SUPPLY & DEMAND

Treasury, not the Fed, may be the catalyst

Hana Securities attributes the reversals in 1994 and 2023 partly to market-position unwinds and issuance adjustments (although the 2023 issuance change came 13 days after the peak). The firm believes such a supply-side response could temporarily push the 10-year yield below 5%.¹⁰ On August 19, Treasury announced that, effective September 9, it would raise liquidity-support buybacks in the 10–20-year and 20–30-year sectors from a maximum $2bn to at least $4bn per operation.³³ But the January 2000 buyback program aimed to retire debt amid fiscal surpluses. Today Treasury continues issuing new debt to finance deficits.³⁴ The same term, “buyback,” does not imply the same effect on yields.

SMALL SAMPLE

Six observations cannot time a peak to the month

The 2004–06 cycle reached its maximum nearly a year later, albeit just 1bp above its pre-final-hike reading. Hana's 1990 peak occurred during an easing cycle. The events that amplified subsequent yield declines also differed from one episode to the next.

Other investors point to forces outside the Fed's control. In an August 21 analysis, PIMCO's Marc Seidner and Pramol Dhawan attributed higher long-term yields to a combination of rising government debt, heavy AI-related corporate bond issuance, and inflation concerns tied to energy and monetary policy. They saw AI issuance as competing with Treasuries for capital, but not as the primary driver. With inflation expectations broadly anchored, they argued that most of the move had been in real yields. Barring an unexpected downturn, they expected the higher term premium to persist, while also noting that today's higher starting yields offer income to long-term investors.³⁵ In the Wall Street Journal's October survey of economists, only 17 of the 64 respondents who offered a 10-year yield forecast expected it below 5% at year-end; more than half expected it to remain above 5% at least through mid-2027 (as reported in secondary coverage).³⁶

The fiscal figures have grown as well. The FY2026 federal deficit, for the fiscal year ended in September, was approximately $2tn—$217bn more than a year earlier—while net interest costs totaled roughly $1.1tn.³⁷ Slowing housing activity would tend to pull long-term yields down, while the burden of Treasury supply works in the opposite direction.

Overseas bond yields are another risk. Hana Securities cites the 1990 peak as an episode in which rising German and Japanese rates helped lift U.S. long-term yields.¹⁰ The French 10-year yield reached 4.93% on October 19, revisiting levels that the Korea Herald Economy reported as close to the highest since 2002 earlier in October. At that earlier point, the spread over Germany was about 160bp, its widest since 2012.^{30, 38} Hana also notes that 30-year yields in Japan and Germany have risen into the 4% range, increasing competition with U.S. Treasuries for long-duration capital.¹⁰

Q6FOMC · OCTOBER 27–28

Four things to watch at—and around—the FOMC

There will be no new dot plot. More important than whether the Fed holds is how expectations change for the remaining number of hikes and the time rates stay elevated. Judge the market reaction relative to the path priced in before the meeting.

How does the statement describe further hikes?
The September statement said the policy action would support a timelier return of inflation to the Fed's 2% goal.² In the September minutes, most participants judged that one more hike before year-end would likely be appropriate.¹² Watch whether the statement signals another move, whether it suggests one or several more hikes, and whether it leaves the timing open.
Does Chair Warsh change the assessment of financial conditions?
Many participants in the September minutes judged financial conditions supportive of growth despite higher long-term yields.¹² The distinction is between Warsh merely acknowledging the rise in long yields and explicitly revising that assessment. Hana Securities sees a key question: will the Fed treat higher long-term yields as tightening that substitutes for additional policy-rate hikes?¹⁰ Chair Powell attached conditions to that argument on November 1, 2023. Tighter financial conditions would have to persist and not merely reflect expectations of future Fed moves. He was also reluctant to translate the effect into an equivalent number of hikes.¹¹ The 10-year yield had already peaked on October 19 that year, before Powell made those remarks.
First the expected policy path, then the long end
Start with changes in the expected rate path embedded in fed funds futures and overnight index swaps (OIS) around the meeting. Then ask whether the 10-year yield moved more than that path would suggest, and whether the term premium, inflation expectations, or supply-demand dynamics also shifted. Yield-curve slope alone cannot identify the final hike. For example, if the 2-year yield rises 15bp and the 10-year rises 5bp, the curve flattens—even as markets may be pricing stronger near-term tightening (illustrative scenario). Flattening is not proof that the hiking cycle is near its end. On October 19, a rise led by longer maturities produced bear steepening; on October 20, weak housing data accompanied a 2bp rise in the 2-year and a 3bp decline in the 10-year, flattening the curve.^{1, 16}
The earnings and supply calendar around the meeting
At the October 21, 1 p.m. U.S. Eastern auction (2 a.m. October 22 Korea time) of $13bn in 20-year Treasuries, watch the high yield, bid-to-cover ratio, and auction tail relative to pre-auction trading levels as an early test of long-end demand.³⁹ After the U.S. market closes the same day, Meta's Q3 earnings (early October 22 Korea time) will provide clues on AI capital spending, financing plans, and cash flow.^{32, 40} Treasury's November 2 quarterly borrowing estimates will indicate how much financing is needed; its November 4 Quarterly Refunding Announcement (QRA) will set out the coupon-security maturity mix and whether the enlarged long-end buyback program will be maintained or adjusted.^{33, 41} The primary-dealer meeting agenda released by Treasury on October 16, for meetings on October 29–30, sought feedback on how the enlarged buybacks had affected liquidity, market functioning, and dealers' intermediation capacity, as well as possible further improvements.⁴²

How to interpret each FOMC outcomeScenario framework, not probabilities. Compare each outcome with pre-meeting pricing (Kiwoom Securities estimated a more-than-90% chance of at least one hike by December as of Oct. 19).
October 28 outcome	Implication for policy path	Implication for early-October peak hypothesis	What else to watch
Hold · Signals one more hike in 2026	Broadly consistent with expectations for a December hike. Whether it would be the final move—or whether further hikes follow in 2027—would remain unclear.	The October 5 yield of 5.31% remains a possible peak, conditional on December being the final hike.	December dots on 2027 path · Nov. 4 QRA
Hold · Signals several more hikes	Higher terminal rate or longer period at restrictive levels. Reaction may be muted if markets already expect further tightening.	Weakens the hypothesis; shifts the window later.	OIS path · Whether 2- and 10-year yields rise together
Hold · Recognizes long-end yields as tightening	Could reduce the need for additional hikes. Watch whether Warsh applies persistence and policy-expectations qualifications similar to Powell's in 2023.	Supports the hypothesis through lower policy-rate expectations.	Whether term premium rebounds
Unexpected rate hike	Could bring forward a hike expected in December, or mark the start of additional tightening. The date of the last hike and total cumulative tightening are separate questions.	Reassess after the initial market shock.	OIS-implied terminal rate · High-yield spreads · Housing data

What evidence would change the assessment?An interpretive framework, not a statistical forecast; monitor over the weeks after the meeting
Observation	Implication for early-October peak hypothesis	Caveat
Expected terminal rate or duration of elevated rates falls after FOMC	Supports the hypothesis through the policy-expectations channel.	A rising term premium could offset the effect.
Policy expectations are unchanged, but pressure on long yields eases	Suggests a shift in supply-demand dynamics or risk compensation.	Check QRA, buybacks, and market liquidity.
Expectations of more hikes and long-end risk compensation rise together	Weakens the hypothesis.	Hard to defend merely because yields are “close to the previous peak.”
Credit stress spreads from individual issuers across funding markets	Reason to reassess the economic outlook and policy path.	Financial turmoil does not automatically translate into a Treasury rally.

KNOWN

Sept. 16: +25bp to 3.75–4.00%; unanimous 12–0 vote
September dot-plot median: 4.1% at year-end 2026 and 2027 (target-range midpoint)
Of 18 September dot-plot contributors, 16 expected another hike in 2026, including four expecting two (Waller)
No dot plot at the Oct. 28 FOMC; next projection round on Dec. 9
Oct. 20: 10-year yield 5.12%, 2-year 4.79% (Treasury); 2026 10-year high 5.31% on Oct. 5
Treasury long-end buybacks of at least $4bn per operation since Sept. 9; borrowing estimates Nov. 2; QRA Nov. 4
October NAHB index at 29, lowest since June 2012; September Pending Home Sales Index at 66.0, lowest since inception

ESTIMATE

Hana Securities: terminal upper bound 4.50% (December and March hikes); peak window Q4 2026–Q1 2027
Hanwha Investment & Securities: terminal upper bound 4.25% (December hike)
Kiwoom Securities: >90% chance of at least one additional hike by December
Of the 86.7bp rise in the fitted 10-year zero-coupon yield from end-2025 to Oct. 16, 41.5bp came from the term premium (Kim–Wright model; estimates vary across models)
CDS for some hyperscalers near 100bp, based on Eugene Investment & Securities chart readings
Duration-adjusted AI corporate bond issuance equivalent to about 52% of 10-year Treasury issuance (Hanwha)
Applying a historical ±4-month window to hypothetical final-hike dates

UNKNOWN

Date of the final hike
Whether the Oct. 5 yield of 5.31% proves to be this cycle's peak
Whether the Fed hikes or cuts during 2027
November QRA issuance mix and buyback changes
Whether AI credit concerns spread into broader funding markets

Historically, the U.S. 10-year Treasury yield has often peaked close to the Fed's final rate hike. Yet the final hike cannot be established in real time, and this cycle's long-term yields are also being shaped by the supply of Treasuries and demand for capital to finance AI investment. The October FOMC will help clarify expectations for the scale and duration of remaining tightening. Treasury's November 2 borrowing estimates and November 4 QRA will show how officials plan to address long-end supply. Meanwhile, the key question is whether strain in housing and AI credit spreads into other sectors. Only after assessing those three forces can we judge whether 5.31% on October 5 marked the peak.

Review of published forecasts and historical cases ahead of the Oct. 27–28 FOMC · Not a prediction of rate direction or a trading recommendation · Data cutoffs and calculations are documented in the sources below

TERMS · GLOSSARY

Final rate hike: The Fed's last policy-rate increase within a tightening cycle. It can only be identified after subsequent holds or cuts.
Term premium: Compensation for the risks of holding a longer-dated bond. A 10-year yield can be decomposed into the expected path of future short-term rates and a term premium. The premium is not directly observable; models such as Kim–Wright and ACM estimate it, with differing results.
Dot plot (SEP): FOMC participants' projections for the federal funds rate at year-end, plotted as the midpoint of the target range. Published with the Summary of Economic Projections in March, June, September, and December, but not at this October meeting.
OIS: An overnight index swap exchanges a fixed rate for a floating rate linked to overnight interest rates. OIS pricing helps infer market expectations of the policy-rate path.
Fitted 10-year zero-coupon yield: The yield on a hypothetical 10-year Treasury zero-coupon bond, derived from the Treasury yield curve by the Kim–Wright model. It differs slightly from the quoted 10-year constant-maturity Treasury yield.
Neutral rate · Restrictive rate: The neutral rate neither stimulates nor restrains the economy; a restrictive rate sits above it and weighs on demand.
Federal funds futures: Futures contracts based on the effective federal funds rate. Together with OIS, they are used to infer the policy-rate path priced by markets.
Quantitative tightening (QT): A central bank reduces its securities holdings, withdrawing liquidity from the financial system.
Powell pivot: Chair Powell's shift in January 2019 toward patience and a pause in further rate hikes.
Modified duration: An estimate of a bond's price sensitivity to yield changes: approximately the percentage change in price for a 1-percentage-point move in yield, with the opposite sign. A higher duration means greater interest-rate risk.
Hyperscalers · Neocloud providers: Large cloud companies operating data centers at immense scale (such as Amazon, Microsoft and Google), versus newer GPU-intensive cloud providers specializing in AI compute rental (such as CoreWeave).
H.15 10-year yield: The Federal Reserve's daily 10-year constant-maturity Treasury yield. Derived from the yield curve, it may differ from the traded yield or closing quote on a specific Treasury security.
QRA: The U.S. Treasury's Quarterly Refunding Announcement (QRA), setting out planned issuance of coupon-bearing notes and bonds and Treasury bills by maturity.
Liquidity-support buybacks: Treasury purchases outstanding, less-liquid securities in the secondary market to improve market functioning. This is distinct from Federal Reserve purchases under quantitative easing (QE).
Bear steepening: A rise in yields in which longer-term rates increase more than shorter-term rates, steepening the yield curve.
CDS: A credit default swap: its premium, quoted in basis points, generally rises as the market demands more compensation for credit default risk.
High-yield spread: The yield premium of below-investment-grade corporate bonds over Treasuries, widely used to gauge credit risk across the market. Index measures often use option-adjusted spreads.
Bid-to-cover ratio · Auction tail: Bid-to-cover measures total bids relative to the amount offered. The tail is the difference between the auction's high yield and the prevailing when-issued yield immediately before bidding. A low bid-to-cover ratio or a large positive tail can signal weak demand.
RPO: Remaining performance obligations: the portion of contracted revenue not yet recognized in the financial statements, broadly analogous to a revenue backlog.

SOURCES · METHODOLOGY · AS-OF DATES

Published October 21, 2026 · Information cutoff U.S. Treasury yields, housing data, and the S&P 500 are through the U.S. October 20 close (morning of October 21 in Korea). Brokerage research is dated October 12–20, with market readings generally through October 19. Kim–Wright estimates were available through October 16. The October 21 U.S. 20-year Treasury auction and Meta earnings were still upcoming as of publication.

As-of dates	Oct. 20 U.S. 10-year and 2-year yields: Treasury constant-maturity series (same values as Fed H.15). Historical peaks and the 2026 YTD high of 5.31% on Oct. 5: Fed H.15 10-year constant-maturity yields (daily curve-derived observations, not necessarily the closing quotes of individual bonds). French and German 10-year yields: Oct. 19 (iM Securities). Fed target rate and vote: Sept. 16; dot plot: Sept. 16; September minutes released Oct. 7. S&P 500 closes: Oct. 6 and 20. October high-yield spread peak: Hana Securities on Oct. 19 and ICE BofA on Oct. 16. September housing starts and pending home sales released Oct. 20. CDS observations: mid-October (Eugene Investment & Securities chart dated Oct. 20; Reuters Oct. 15). NAHB released Oct. 19; CoreWeave June 30 Form 10-Q. The 20% October hike probability is from Hana's Oct. 12 report.
Snowshagal calculations	Cycle peak = highest H.15 10-year yield from the first-hike decision through two years after the final hike (ties use earliest date). Window sensitivity: using one year after the final hike leaves all six peaks unchanged; a six-month limit changes the 2004–06 peak to 5.25% on June 26, 2006 (−0.1 months). Month offset = (peak date − final-hike FOMC decision date) ÷ 30.44 days: 1994 −2.8; 1997 +0.6; 2000 −3.8; 2007 +11.4 (June 26 and 28, 2006 readings of 5.25% = −0.1); 2018 −1.3; 2023 +2.8. Time from first hike to yield peak: 9.1 months; 17 days; 6.7, 35.4, 34.8 and 19.1 months, respectively. Cumulative hikes by peak: 175, 25, 75, 425, 200 and 525bp (the five-cycle mean excluding 1997 is 21.0 months and 280bp, matching Hana). Two-year change = month-end yield in the corresponding month two years later minus peak: 1994 6.06 − 8.05 = −199bp; 1997 5.36 − 6.98 = −162bp; 2000 5.07 − 6.79 = −172bp; 2007 3.53 − 5.26 = −173bp; 2018 0.84 − 3.24 = −240bp; 2023 4.11 − 4.98 = −87bp. Conditional ±4-month calendar bands: for Dec. 9, Aug. 9, 2026–Apr. 9, 2027; for Mar. 17, Nov. 17, 2026–Jul. 17, 2027. Ten-year yield minus upper target bound = 5.12 − 4.00 = 112bp. YTD changes, from Treasury rates on Dec. 31, 2025: 10-year 5.12 − 4.18 = +94bp; 2-year 4.79 − 3.47 = +132bp. Oct. 5–20 10-year decline = 5.31 − 5.12 = 19bp. Kim–Wright decomposition (Dec. 31, 2025–Oct. 16, 2026): fitted 10-year zero-coupon yield 4.2401% → 5.1066% (+86.7bp); term premium 0.5710% → 0.9857% (+41.5bp); implied change in expected short-rate path +45.2bp. Oct. 5–16: fitted yield 5.2845% → 5.1066% (−17.8bp); term premium 1.0965% → 0.9857% (−11.1bp, or 62% of the decline). France–Germany spread = 4.9309 − 3.4776 = 145.3bp. S&P 500 return = 7,712.57 ÷ 7,818.93 − 1 = −1.4%. CoreWeave Q2 net interest expense/revenue = $640m ÷ $2,575m = 24.9%. CoreWeave June 30 borrowings = recourse $6,235m + $25,170m and nonrecourse $1,278m + $2,385m = $35,068m, summed from balance-sheet carrying amounts.
Estimates read from charts	The “around 100bp” CDS levels for selected hyperscalers are based on Eugene Investment & Securities' figures and commentary. Reuters reported record-high trading levels for Amazon, Alphabet, Meta and Nvidia without providing an individual CDS quote for each.
Differences among data sources	Oct. 19 10-year: Kiwoom Securities 5.140% (+97.3bp YTD); iM Securities 5.1418%; Treasury and Fed H.15 5.15%. Oct. 19 2-year: Kiwoom 4.772%; Treasury/H.15 4.77%. Reports give Oct. 20 10-year yields of 5.09–5.12%; this note uses the Treasury's 5.12%. German 10-year: Kiwoom 3.456%; iM 3.4776%. The 2026 10-year high: Fed H.15 5.31% on Oct. 5; iM 5.3046%; one report cited an Oct. 7 intraday reading of 5.36%. France–Germany spread: roughly 160bp in early October (reporting), versus 145bp on Oct. 19 (calculation). The 2000 peak of 6.79% was recorded on both Jan. 20 and 21; the 1997 peak of 6.98% on both April 11 and 14.
Checks against the cited research	Hana Securities' summary says that excluding 1990, five peaks occurred within four months of the final hike, but its own body text treats 2007 as an exception, leaving four out of five. Hana's “six peaks” include 1990 (an easing period) and exclude 1997, so this note instead counts all six Fed tightening cycles. Hana's Chart 2 dates the 2023 peak “2023/01,” at odds with its body text and Chart 19 (October). Its 2007 peak also varies within the report—5.26% in Chart 2, 5.29% in the text, and 5.2971% in Chart 12—as do its 1994 readings (8.05% in text versus 8.014% in Chart 6, dated October). This note uses the Fed's daily H.15 yields throughout.
Estimates and interpretation	Terminal-rate and peak-window projections belong to the respective brokerages. The ±4-month bands are conditional illustrations of a historical observation, not forecasts. The Kim–Wright decomposition is model-based and differs from ACM and other models. “Stress is evident in housing and some AI firms” is this note's interpretation of the dashboard indicators. The scenario and assessment tables are interpretive tools, not assigned probabilities.

U.S. Treasury, “Daily Treasury Par Yield Curve Rates.” Dec. 31, 2025: 2-year 3.47%, 10-year 4.18%; Oct. 19, 2026: 2-year 4.77%, 10-year 5.15%; Oct. 20, 2026: 2-year 4.79%, 10-year 5.12%, 30-year 5.42%. home.treasury.gov
Federal Reserve, “Federal Reserve Issues FOMC Statement,” Sept. 16, 2026. “raise the target range for the federal funds rate by 1/4 percentage point to 3-3/4 to 4 percent”; 12–0 vote; “Today's policy action will support a timelier return to the Committee's 2 percent goal.” federalreserve.gov The first hike since 2023 is also documented in MUFG and Trading Economics reporting and the Fed's “Open Market Operations” history.
Federal Reserve, “Summary of Economic Projections,” Sept. 16, 2026. Median fed funds rate: 4.1% in 2026, 4.1% in 2027, 3.9% in 2028 and 3.2% longer run; 2026 PCE inflation projection 3.7%, core PCE 3.4%. federalreserve.gov
Federal Reserve, “Meeting Calendars and Information.” FOMC meetings Oct. 27–28 and Dec. 8–9, 2026 (with SEP); tentative 2027 meetings include Jan. 26–27 and Mar. 16–17 (with SEP). federalreserve.gov Decision scheduled for 2 p.m. U.S. Eastern on Oct. 28 = 3 a.m. Korea time on Oct. 29.
Federal Reserve H.15, “Market Yield on U.S. Treasury Securities at 10-Year Constant Maturity, Quoted on an Investment Basis” (FRED DGS10, daily observations derived from the yield curve; through Oct. 19, 2026). Peaks: Nov. 7, 1994, 8.05%; Apr. 11/14, 1997, 6.98%; Jan. 20/21, 2000, 6.79%; June 12, 2007, 5.26% (June 26/28, 2006, 5.25%); Nov. 8, 2018, 3.24%; Oct. 19, 2023, 4.98%. Hana Securities also lists a May 2, 1990 peak of 9.09%. 2026 high: Oct. 5, 5.31%; Oct. 19, 5.15% (2-year 4.77%). Corresponding month-end yields two years later: Nov. 1996, 6.06%; Apr. 1999, 5.36%; Jan. 2002, 5.07%; June 2009, 3.53%; Nov. 2020, 0.84%; Oct. 2025, 4.11%. fred.stlouisfed.org
Federal Reserve, “Open Market Operations Archive.” Target changes in 1994–2000: first hike Feb. 4, 1994, through a 75bp hike on Nov. 15; a 50bp hike on Feb. 1, 1995, to 6.00%. March 25, 1997: +25bp to 5.50%, the only hike after 75bp of cuts in 1995–96. First hike June 30, 1999, through a 50bp hike on May 16, 2000, to 6.50%. federalreserve.gov
Federal Reserve FOMC statement, June 29, 2006: 25bp hike to 5.25%. federalreserve.gov
Federal Reserve FOMC statement, Dec. 19, 2018: increase to 2.25–2.50%. federalreserve.gov
Federal Reserve, “Open Market Operations”: hikes from Mar. 17, 2022, through July 27, 2023 (effective dates; FOMC decisions were the previous day); Sept. 17, 2026, effective target of 3.75–4.00%. federalreserve.gov
Hana Securities, Heo Seong-woo, “When Will U.S. Rates Peak?” (Global Fixed Income), Oct. 20, 2026. P. 1 summary: all six major peaks described as reversals due to high-rate side effects; average change two years later −164bp; first hike to peak average 21 months/280bp; excluding 1990, five claimed within four months of final hike; 2007 described as an exception after a year on hold; terminal rate 4.50% with hikes in December and March; yield peak likely Q4 2026–Q1 2027, with risks from a higher-for-longer (H4L) December dot plot or earlier strain. P. 2: the 1997 parallel, policy rate–10-year yield gap around 110bp/max. 130bp versus historical peak average of 100bp; 1994/2023 supply-demand responses; possibility of a temporary dip below 5%. P. 3 Chart 2: six peaks and triggers, citing Lance Roberts. P. 8: 2007 analogy; risks in private credit, AI issuance and commercial real estate. P. 9 box: Greenspan conundrum versus 2026, term premium reportedly highest in 12 years, Japanese/German 30-year yields around 4%. P. 12: November 2023 QRA and Powell remarks; upcoming QRA and Chair Warsh's remarks as catalysts.
Federal Reserve, Chair Powell's press conference, Nov. 1, 2023. “the tighter conditions would need to be persistent”; “they can't simply be a reflection of … expected policy moves from us.” Powell was cautious about expressing the impact as an equivalent number of hikes. Treasury's QRA was released the same day. federalreserve.gov
Federal Reserve, “Minutes of the Federal Open Market Committee, September 15–16, 2026,” released Oct. 7. “Participants generally assessed inflation risk as skewed to the upside”; “labor market conditions were stable”; “Several participants stated that they viewed the current policy rate as not restrictive or only mildly restrictive”; “Many participants commented that, despite the recent rise in longer-term Treasury yields, financial conditions appeared to be supportive of economic growth”; “A few participants commented that housing was a sector in which financial conditions did not appear supportive of activity”; “most participants assessed that another increase … would likely be appropriate by year end.” The Desk briefing attributed some yield increases to a higher anticipated rate path and strong data; market commentary pointed to geopolitical concerns, buyback uncertainty, and heavy private debt issuance for AI infrastructure as drivers of higher term premia. federalreserve.gov
Federal Reserve Board, Kim and Wright (2005), three-factor model; FRED THREEFYTP10 (10-year zero-coupon term premium) and THREEFY10 (fitted 10-year zero-coupon yield). Dec. 31, 2025: yield 4.2401%, premium 0.5710%; Oct. 5, 2026: yield 5.2845%, premium 1.0965% (premium highest since Apr. 14, 2010: 1.0975%); Oct. 16, 2026: yield 5.1066%, premium 0.9857% (latest available in the report). fred.stlouisfed.org
Federal Reserve, Governor Christopher Waller, “The Signaling Value of the Summary of Economic Projections,” Istanbul Economic Forum, Oct. 8, 2026. “Sixteen of the 18 participants who submitted dots anticipated at least one more hike”; “Four of those 16 expected two additional hikes”; “The hikes do not need to come at consecutive meetings.” The 2027 dots “may reflect hikes early in the year and then cuts later in the year”; “eight participants expect that rates will be 50 basis points higher at the end of 2027 than they are today.” Market-implied probability: nearly 80% of at least two hikes by the March 2027 meeting, 33% of three or more. federalreserve.gov
Hanwha Investment & Securities Research Center, “2027 Annual Outlook: Silver Lining,” Oct. 19, 2026. P. 38: 2026 AI-sector bond issuance converted to $187.1bn in 10-year Treasury equivalents, about 52% of 10-year Treasury issuance over the same period. Uses the Dallas Fed's duration-equivalent approach: issue size × modified duration ÷ 8 (assumed 10-year Treasury modified duration), based on issue dates. Forecast terminal target-range upper bound: 4.25%, following one more December hike and a hold thereafter.
Kiwoom Securities, “KIWOOM FICC DAILY,” Oct. 20, 2026. Oct. 19 U.S. Treasuries: 2-year 4.772%, 10-year 5.140% (daily change +6.69bp, weekly −9.81bp); German 10-year 3.456%; “bear steepening led by longer maturities”; French 10-year rose more than 14bp; “the probability of additional tightening by December exceeds 90%” (interpreted here as at least one hike by the December meeting).
Hana Securities, Park Jun-woo and Heo Seong-woo, “[Third Week of October] How Far Must Stocks Fall Before Yields Turn?” (Weekly Hana Fixed Income), Oct. 12, 2026. October hike probability previously near 70%, falling to around 20% amid Waller's caution on consecutive hikes and other factors. Rising yields often stalled or reversed after an 8% S&P 500 drawdown. September FOMC minutes: most participants favored another hike before year-end.
Trading Economics, “Warsh Comments Boost Bets on Fed Rate Hold in October,” Oct. 16, 2026. At an IMF event in Bangkok, Warsh said, “I don't prejudge meetings.” Expectations of an October hike declined. tradingeconomics.com
Eugene Investment & Securities, Park Jae-hwan, “Reassessing the AI Demand Landscape” (Global Tech Weekly), Oct. 20, 2026. P. 1: CoreWeave's April bond yield above 15% (9.75% coupon), CDS above 1,000bp, interest expense about 25% of revenue, over $100bn RPO, and an assessment of limited practical bankruptcy risk. P. 3 chart: some hyperscaler CDS near 100bp; Oracle presented separately.
Reuters, Tom Westbrook and Amanda Cooper, “AI Credit Fears Push Tech Default Insurance to Record Highs,” Oct. 15, 2026 (republished by BusinessDay). Record-high CDS trading levels for Amazon, Alphabet, Meta and Nvidia; Oracle five-year CDS at 288bp versus 260bp a week earlier; CoreWeave above 1,000bp. businessday.co.za
CoreWeave Form 10-Q for quarter ended June 30, 2026. RPO $103.7bn: 41% recognized over next 24 months, 39% in months 25–48, remainder in months 49–78. Q2 revenue $2.575bn; net interest expense $640m. Balance-sheet borrowings (recourse/nonrecourse; current/noncurrent) total about $35.07bn. sec.gov
NAHB Eye on Housing, Robert Dietz, “Builder Sentiment Falls on Rising Rates,” Oct. 19, 2026. “fell three points to 29 in October”; “The reading was the lowest since June 2012.” Mortgage rates recently well above 7%. eyeonhousing.org
National Association of Realtors (NAR), “Pending Home Sales,” Oct. 20, 2026. September Pending Home Sales Index: 66.0, down 7.3% month over month and 11.3% year over year; all four regions declined; lowest in the series (base year 2001 = 100). nar.realtor · HousingWire 「NAR pending home sales index falls to record low in September」 2026.10.20 · housingwire.com
U.S. Census Bureau, “Monthly New Residential Construction, September 2026,” released Oct. 20, 2026 at 8:30 a.m. ET. Housing starts: seasonally adjusted annual rate of 1,221,000 (−5.2% versus revised August 1,288,000; 90% confidence interval ±11.8%). Building permits 1,400,000 (−0.2%); single-family starts 859,000 (−4.6%). census.gov
Freddie Mac Primary Mortgage Market Survey (FRED MORTGAGE30US). The 30-year fixed rate was 7.40% on both Oct. 8 and Oct. 15, 2026; the last higher reading was 7.44% on Nov. 16, 2023. fred.stlouisfed.org
Hana Securities, Lee Jae-man, “What to Ask After the KOSPI Selloff” (Equity Strategy), Oct. 19, 2026. U.S. high-yield credit spread widened from about 260bp in late August to an October high of 331bp.
ICE BofA U.S. High Yield Index Option-Adjusted Spread (FRED BAMLH0A0HYM2). Aug. 31, 2026: 2.63%; Oct. 16: 3.31% (highest since August); Oct. 19: 3.29%. fred.stlouisfed.org
iM Securities Credit Research, “Bull Flattening Alone May Not Sustain Credit Strength,” Oct. 19, 2026. U.S. high-yield spreads remained below their median of 356bp since 2019.
S&P 500 close on Oct. 6: 7,818.93, a record closing high (the prior day's close reported by CryptoDaily on Oct. 7). cryptodaily.co.uk An Oct. 6 report titled “S&P 500 Notches Record Closing High” (roic.ai). roic.ai · 10/19 7,690.73, 10/20 7,712.57(FRED SP500) · fred.stlouisfed.org
iM Securities, Park Sang-hyun, “The Headwinds Persist, but They Are Not Intensifying” (Economy Brief), Oct. 20, 2026. U.S. 10-year high 5.3046%; then 5.1418%. French 10-year on Oct. 19: 4.9309%; German 10-year: 3.4776%. Although CDS rose for CoreWeave and hyperscalers, U.S. credit spreads were stable, and broad contagion was not yet apparent.
IMF Blog, Athanasios Vamvakidis and Jason Wu, “Global Financial Markets Face Rising Risks From AI, Bond Yields, and Cyberattacks,” Oct. 13, 2026, marking publication of the October Global Financial Stability Report. Fourteen publicly disclosed AI-related deals since 2023: roughly $300bn in equity investment, $650bn in debt, and $1.4tn in future revenue commitments. imf.org
Kiwoom Securities, Han Ji-young, “Breaking the Negative Cycle Around AI Stocks” (Investment Strategy), Oct. 20, 2026. Oct. 19 U.S. Treasury yields: 10-year 5.140% (daily +6.7bp; YTD +97.3bp), 2-year 4.772% (YTD +129.9bp), 30-year 5.467%. S&P 500 at 7,690.73. Cited feedback loop: “AI data-center investment expands → external debt grows → high rates persist and credit spreads widen → new-project and refinancing costs rise → payback periods lengthen.” Big Tech earnings, starting with Meta, could be a turning point.
U.S. Treasury, “Treasury Announces Increased Sizes of Nominal Long-End Liquidity Support Buybacks,” Aug. 19, 2026. Liquidity-support buybacks in the 10–20-year and 20–30-year nominal coupon sectors raised from a $2bn maximum to at least $4bn per operation, effective Sept. 9; next-quarter guidance due with Nov. 4 QRA. govdelivery.com (U.S. Treasury distribution)
U.S. Treasury, “Statement of Treasury Secretary Lawrence Summers” (LS-330), Jan. 13, 2000. “we expect to buy back as much as $30 billion this year.” FY1999 fiscal surplus $123bn; $140bn in privately held Treasury debt retired over two years. home.treasury.gov
PIMCO, Marc Seidner and Pramol Dhawan, “What's Pushing Long-Term Bond Yields Higher?,” Aug. 21, 2026. Drivers include rising sovereign debt, heavy AI corporate issuance, and inflation concerns from energy and monetary policy; AI issuance was not their main driver. “The recent volatility has largely been a real-yield event”; “Barring an unexpected economic downturn, we expect this higher term premium to endure.” pimco.com
Wall Street Journal October economists' survey (Oct. 2–7; 71 total respondents), cited in secondary reporting. IndexBox, Oct. 20, 2026: of 64 respondents to the 10-year yield question, only 17 expected a sub-5% yield at end-2026; more than half expected it above 5% through at least mid-2027. indexbox.io CRE Daily also cited a 5.1% year-end 10-year yield forecast. credaily.com
American Action Forum, “U.S. Treasury: FY 2026 Deficit Totaled $2 Trillion,” Oct. 18, 2026. Based on Treasury's September Monthly Treasury Statement: FY2026 deficit about $2tn (up $217bn, or 12%, year over year); net interest costs around $1.1tn. americanactionforum.org
Herald Economy (Korea), “As U.S. and French Bonds Lose Favor, Germany Gains: A New Safe-Haven Hierarchy,” Oct. 8, 2026. Oct. 7 U.S. 10-year intraday high 5.36%; French 10-year briefly 4.93%, near its highest since 2002; France–Germany spread about 160bp, widest since 2012. heraldk.com
U.S. Treasury auction announcement (Oct. 15, 2026; republished by Newsquawk). $13bn of 20-year Treasuries to be auctioned Oct. 21 and settled Oct. 22; $26bn of five-year TIPS to be auctioned Oct. 22. newsquawk.com
Meta, “Meta to Announce Third Quarter 2026 Results.” Earnings after the U.S. market close on Wednesday, Oct. 21, 2026; conference call at 4:30 p.m. ET = 5:30 a.m. Korea time on Oct. 22. investor.atmeta.com
U.S. Treasury, “Most Recent Quarterly Refunding Documents.” Primary-dealer meeting agenda published at noon Oct. 16, 2026; next financing estimates scheduled Nov. 2; next quarterly refunding announcement Nov. 4. home.treasury.gov
U.S. Treasury, “Dealer Agenda November 2026” (November QRA dealer questions), published at noon Oct. 16, 2026; responses due Oct. 26, dealer meetings Oct. 29–30. Question 3 asks how the expanded long-end buybacks have affected liquidity, market functioning, and dealer intermediation capacity, and what further improvements might help. home.treasury.gov

The U.S. 10-Year Yield Tends to PeakNear the Final Rate Hike