Ninety Days From Today

Count it out on a calendar. Ninety days from this morning is December 30, a Wednesday, which means this window closes two days before the year does. Between here and there sit two Federal Reserve meetings, on October 27 and 28 and again on December 8 and 9; a midterm election on November 3; a Treasury quarterly refunding on November 4; a government funding deadline on December 11; a Thanksgiving on November 26; and a Christmas on December 25. That is the window this article is about: one quarter of a year in which the world’s largest bond market has to price a war in its eighth month and a central bank that has told it, in plain language, to expect more.

Two weeks ago this series said the ten-year would land somewhere between 5% and 5.4% by December 17, and it named the de-escalation branch as “the single largest risk to the bearish case.” Both halves of that deserve an answer, and the honest answer is that the level call arrived seven weeks early while the risk it named turned out to be the wrong risk.

The ten-year did not wait for December. It closed September 30 at 5.29%, and by the morning of October 1 it had touched 5.34%, its highest level since early 2002 (Reuters; MarketWatch). The marker this series has been tracking since February, the 5.041% intraday print from mid-September that was itself the highest since 2007, is now a quarter of a point in the past. That is not a forecast arriving early. That is a forecast being overtaken.

And the peace that was supposed to knock it back did not come. Talks stalled this week, and on Tuesday Secretary of State Marco Rubio ordered Iran’s delegation to the United Nations, including Foreign Minister Abbas Araghchi, to leave the country; the delegation flew to Doha before dawn (Axios). President Trump described the choice to reporters as “blow them up or make a deal,” and then added: “Maybe I blow ’em up” (Politico). Iran had offered a seven-day roadmap to reopen the Strait of Hormuz in exchange for the naval blockade being lifted and its frozen assets released. The offer was refused (Al Jazeera).

So here is the thing this installment exists to explain, because it is the most interesting fact in the market right now. Oil fell anyway.

Brent settled at $102.59 on September 29 and traded near $104 on October 1, which is roughly where it sat when the last installment went to press, and it is up about 13% on the month and about 42% from its late-June low (Reuters; FXStreet). It did not spike on the collapse of the talks. And the reason it did not is that the strait has been quietly reopening without a deal. Tanker transits through Hormuz are approaching prewar levels. Kpler puts Middle East crude exports at about 16.3 million barrels a day in September, the highest since the war began, against a prewar figure near 19.5 million. JPMorgan’s estimate is higher, near 17.5 million barrels a day, which is about 98% of prewar (El Pais).

The barrels are moving again. The price is not coming down. That gap is the whole argument of the next ninety days, and it is a harder problem for the bond market than a war that simply continued.

Where the Market Stands Today

The numbers as of the last completed session, with the October 1 open noted where it has already moved.

  • The 10-year Treasury closed at 5.29% on September 30, up from 5.26% on September 29 and 5.01% on the day of the Fed’s hike. It touched 5.34% on October 1, its highest level since early 2002 (U.S. Treasury daily par yield curve; Reuters).
  • The 30-year Treasury closed at 5.64%, and reached 5.67% on October 1, with an intraday high of 5.683%. That is the highest since 2002 (Bloomberg Law).
  • The 2-year Treasury, the maturity most sensitive to what the Fed does next, closed at 4.88%. It has moved fourteen basis points since the September 16 hike. The ten-year has moved twenty-eight (U.S. Treasury).
  • The 10s30s spread closed at 35 basis points, against 34 on the day of the hike, 32 on September 25, and 48 on September 3. It has spent the entire month inside a fourteen-point band while the level of the curve rose more than half a percentage point (FRED).
  • The 2s10s spread closed at 41 basis points, up from 27 on the day of the hike. The belly of the curve is where the shape actually changed (U.S. Treasury).
  • Brent crude settled at $102.59 on September 29 and traded near $104 on October 1. WTI sat near $92.60. The gap between paper and physical has become the story: cargoes for immediate delivery have changed hands near $121 a barrel, per Argus, and Gulf-to-East-Asia shipping costs have run near $34 a barrel against roughly $3 in January (Business Insider; The New York Times).
  • The Strategic Petroleum Reserve sits at its lowest level since 1982, after more than 130 million barrels were withdrawn since the war began on February 28 (The New York Times).
  • Refined products are the bottleneck that did not clear. Gulf exports of gasoline, diesel, and jet fuel remain near half their 2025 average because of damaged refining capacity, and retail diesel has set another record near $6.50 a gallon (AGBI; The Hill).
  • The average 30-year mortgage is 7.03%, up from 6.95% the week before and 6.30% a year ago, per Freddie Mac (Freddie Mac).
  • The federal funds target range is 3.75% to 4%. The Fed’s September projections put the median year-end 2026 rate at 4.1%, implying one more quarter-point move, and 16 of 18 participants expect at least one more hike this year (Federal Reserve SEP).
  • August PCE inflation came in softer than expected: headline 3.4% year over year against a 3.7% estimate, and core 3.0% against 3.3%. July core was revised down to 3.0% from 3.3% (FXStreet).
  • That soft print moved the next hike rather than canceling it. Odds of a move at the October 27 and 28 meeting fell to roughly 35%, while the December 8 and 9 meeting is now priced between 70% and 100% depending on which market you ask (MacroOdds; Polymarket; CME via FXStreet).
  • ADP reported private payrolls up 90,000 in September, against a consensus near 68,000, while second-quarter GDP was revised up to 2.2% (Reuters).
  • The University of Michigan’s final September sentiment index came in at 48.1, down from 51.7 in August and 55.1 a year ago, a four-month low. Year-ahead inflation expectations jumped to 4.6% from 4.0% (University of Michigan; The Hill).
  • And the backdrop: the 10-year yield has now risen for seven consecutive months, the first such streak since 2011. Germany’s 10-year bund reached 3.62%, its highest since 2008. Japan’s 10-year reached 3.126%, its highest in three decades (FXStreet).

Read the third and fourth bullets together. The two-year has barely moved since the Fed hiked, because the front end has already priced the end of the tightening cycle. The long end has moved twice as far, because it is pricing a question the front end does not have to answer. That difference is the subject of the section after next.

The Strait Is Reopening and the Price Is Not Falling

Most people’s instinct about wartime bonds was formed in 1990, and this series has argued for eight months that the instinct does not apply, because the Gulf War was a demand shock and this is a supply shock. That argument has held. What has happened this month is subtler and worse.

A supply shock that is caused by a chokepoint prices the chokepoint. Open the chokepoint and the price falls. That is the relief trade the market has been waiting for since February, and on the evidence of the last four weeks, it is not going to arrive on schedule, because the constraint has moved downstream while everyone watched the water.

Look at what recovered and what did not. Barrel flows recovered, through workarounds: pipelines that avoid the strait entirely now carry about 40% of Gulf exports, against 17% before the war, and roughly two-thirds of oil volume moves by ship-to-ship transfer (AGBI). Refined products did not recover, because refineries were damaged and rebuilding them is a construction project rather than a reopening. Diesel, the fuel that moves everything by truck and train, is at a record. Insurance did not recover, because the underwriting question is not whether the strait is open but whether it is safe, and three vessels were struck by projectiles on September 29 alone, the most in a single day since July (World Ports Organization). Inventories did not recover, because the reserve that would have buffered a second disruption has been drawn down to a 1982 low.

JPMorgan’s Natasha Kaneva made the point in a sentence worth keeping: higher crossings reflect the industry’s “capacity to operate under sustained risk,” not improved safety (Business Insider). That distinction is the entire reason the price has not fallen, and it has a direct consequence for the bond market that the peace branch never considered.

If the inflation impulse were a chokepoint, then a ceasefire would end it, and the Fed’s problem would solve itself without the Fed doing anything. That was the optimistic version of the de-escalation branch in the last installment, and it is why that branch mattered so much: the ten-year could give back fifty basis points in a fortnight without a single policy change. But if the impulse is refining capacity, insurance premia, and a depleted buffer, then a ceasefire does not end it. It changes the headline without changing the arithmetic. The war can stop and the inflation can continue, and the Fed would still have to finish the job it started.

Notice what the official forecasts assume. Goldman Sachs’ base case has Brent easing to $85 by the end of 2026 and $80 in 2027, contingent on no renewed escalation (Business Insider). That is a forecast of relief. The physical market, where a barrel for immediate delivery is worth $121 and getting it from the Gulf to Asia costs $34, is not behaving as though relief is coming. When the paper market and the physical market disagree this sharply, the physical market is usually the one that knows something.

This is what the long end has been saying all month, and it is why the selloff has been global rather than American. The war is the accelerant. The fuel is that governments everywhere are trying to fund large deficits into a market that has stopped assuming the inflation will pass.

What the 10s30s Spread Did, and Where the Signal Moved

This section has to hold itself accountable, because the last installment named two branches for the spread between the ten-year and the thirty-year and the market chose a third.

The prediction was this. Either the spread re-widens toward its historical average as the long end wakes up to the supply story, which I called the vigilante scenario, or it keeps compressing, which I called the complacent reading and gave a short shelf life. It was 34 basis points on the day of the hike, already the tightest of 2026, against an average near 55 from 1990 through 2019.

It compressed further, to 32 basis points on September 25. Then it drifted back out to 35 on September 30 and stopped. Over the same three weeks the ten-year rose from 5.01% to 5.29% and the thirty-year from 5.35% to 5.64%.

So the market did neither of the things I described. It did something more informative. The vigilante arrived in level rather than in shape. The judgment landed on the whole curve at once, lifting it by nearly half a percentage point, while the gap between the ten and the thirty stayed pinned in the mid-thirties. That is not the long end waking up and demanding compensation relative to the belly. That is every maturity being repriced together, which is what happens when the market revises its view of the entire path of inflation rather than its view of the distant future specifically.

And the shape did move, just not where I was looking. The 2s10s spread went from 27 basis points on the day of the hike to 41 by the end of September. That is a bear steepening at the belly, and it is the cleanest read on the committee’s problem that the market offers. The two-year is anchored because traders believe the hiking cycle is nearly over: the Fed’s own dots say one more move and then a long pause, and the soft August PCE print strengthened that reading. The ten-year is climbing because the market believes the pause will happen at a rate that does not beat inflation, and that the government will keep issuing into the result.

Put plainly: the front end has stopped arguing about whether the Fed will win, and the long end has started arguing about what winning costs. Those are different questions, and the reason they are showing up in the 2s10s rather than the 10s30s is that the ten and the thirty are now being repriced by the same force. Both are long enough to be hurt by an inflation impulse that outlasts the war, and both are far enough out that the difference between ten years and thirty years has stopped being the interesting variable.

For the next ninety days the practical translation is this. The 10s30s in the mid-thirties should no longer be read as a signal of anything. It compressed because the Fed’s hike removed the argument for a long-end bid based on imminent cuts, and it has now settled at a level that simply reflects a market pricing one inflation regime across all long maturities. The instrument to watch this quarter is the 2s10s. If it keeps steepening while the funds rate stays put, the market is telling you it expects the Fed to stop before it wins. If it flattens back toward 27, the market is telling you it expects the Fed to keep going. Everything else in this article follows from which of those two happens.

The Buyback Scorecard, Third Entry

Two installments ago this section had an asterisk beside it, because the Treasury had doubled its long-end buyback operations but the larger program had not yet run. One installment ago the asterisk came off, and the verdict was that the program had failed on its own terms: the ten-year rose from 4.83% to 5.01% over the first week and a half of the bigger operations. That verdict stands, and this month the program did something worse than fail. It could not fill its own orders.

Here is the arithmetic, from Treasury’s own operation reports. The enlarged program, announced August 19 and running from September 9 through the November 4 refunding, raised the maximum size per operation to at least $4 billion for the ten-to-twenty-year sector and $6 billion for the twenty-to-thirty-year sector (U.S. Treasury).

  • On September 17, Treasury offered to buy up to $4 billion of bonds maturing between 2033 and 2036. Dealers offered $9.74 billion. Treasury accepted $2.385 billion, about 60% of its own cap, across 6 of the 10 eligible issues (U.S. Treasury buyback results).
  • On September 24, the twenty-to-thirty-year operation had a $6 billion cap. Dealers offered $10.489 billion. Treasury accepted $4.078 billion, about 68% of the cap, across 12 of 35 eligible issues. The operation before it had taken 86% of its cap (Caproasia).

Two readings follow, and they are both unflattering.

The first is that the department has quietly decided the long end is not worth supporting at these prices. Announcing a larger bid is free. Paying it is not, and at 5.6% on the thirty-year, every bond the Treasury retires is a bond it may have to reissue later at a higher coupon. Raising the cap in public and then taking two-thirds of it is what a buyer does when the buyer wants the announcement more than the bonds.

The second reading is more cynical and harder to dismiss. If the goal were genuinely liquidity, the natural measure would be whether trading got easier, and the answer appears to be that trading was never difficult. Thomas Simons of Jefferies told Reuters that holders “don’t clearly need the liquidity that bad” (Reuters via Investing.com). And the bonds the program actually targets have a second property that has nothing to do with liquidity: many are low-coupon securities issued during the pandemic, now trading well below face value, which makes them cheap to retire and a quiet way to manage the maturity profile rather than to support the market (Reuters via Investing.com).

Padhraic Garvey of ING offered the most generous reading available, which is that the swap spread narrowed and that this is “all he can do” (Reuters via Investing.com). Take that at face value and the program is working as intended and simply does not matter very much. Take the other reading and the program is a maturity-management tool wearing a liquidity label. Either way, the answer to the question this section has been asking since August is now clear: buying your own bonds does not lower the rate at which the market will lend to you for thirty years.

Watch November 4. That is the date the enlarged program expires, the date of the quarterly refunding, and the date Treasury has said it will tell the market what future buyback sizes will be. It is also the morning after the midterm election. If the cap comes down or the operations go unfilled again, the department has conceded that the experiment is over. If the cap goes up, the market will read it within minutes as an admission that the long end still needs a buyer of last resort (U.S. Treasury).

The Governance Question

This series treats the rule of law as a standing part of the bond market rather than a political aside, because the term premium is where governance is priced. Four threads ran this month, and the newest one is the one that should worry a bondholder most.

The statistical foundation under the Fed’s decisions is being rewired. On September 23, Senator Elizabeth Warren wrote to Chair Kevin Warsh about a Commerce Department directive issued August 19 that removed the ban on political interference from its scientific integrity policy, the rulebook that governs how the Census Bureau and the Bureau of Economic Analysis produce their numbers (U.S. Senate Banking Committee). This is the sharpest entry in this section since the series began, and it is not a partisan point: the PCE inflation figure the Fed used on September 30 to justify a hike is produced under that policy. A central bank that sets the price of money by reading numbers whose production rules were just loosened is a central bank whose decisions are harder to trust, and the market charges for the difficulty.

The tariff refunds have become an administrative scramble. The Supreme Court ruled in February that the emergency statute does not authorize tariffs and left the remedy to the Court of International Trade, which ordered broad refunds. Customs has certified roughly $134.7 billion in potential refunds and sent about $122 billion back to Treasury, and on October 6 it opens the next phase of the process, covering about $11.4 billion in older entries that had already been closed out (Yahoo Finance; The Financial Wire). The Department of Justice is appealing the breadth of the refund order on the argument that only the named plaintiffs are entitled to relief (Ice Miller via JD Supra). A transfer of that size, resolved by who filed in time and who did not, is a corruption risk in plain sight, and the statute of limitations does not run until early 2027.

The exemption channel now has an academic literature. A study published in the Journal of Financial and Quantitative Analysis found that campaign contributions to the party in power raised the odds a tariff exemption would be granted, that contributions to the opposition lowered them, and that each approval was worth a median of about $51 million in firm value (Journal of Financial and Quantitative Analysis). That study predates this administration; it describes a mechanism, not this month’s news. But it is the quantitative form of the accusation Senate Democrats have been making since February, that the exemption process “appears to favor the politically connected,” and it is what a bond market means when it says it is charging for discretion (NPR).

And the pressure on the Fed has not let up, even though the Fed won. The September 16 hike was unanimous, 12-0, against the president’s public demand for cuts to 1% (CNN). The institutional read is genuinely reassuring, and it should be credited. The structural read is that the administration restarted its effort to remove Governor Lisa Cook in August after the Supreme Court blocked the first attempt in June, and that a president who describes independent votes as a personal attack has told the market what he believes the institution is for (SCOTUSblog; Reuters). Investors price the risk that he eventually gets his way. That is a basis point cost no hawkish statement can refund.

The Ninety-Day Map

Take the premise seriously: no improvement in the war, no breakthrough, no formal reopening of the strait. Here is the path from October 1 to December 30.

October 2. The September employment report. Consensus sits near 84,000 jobs with unemployment steady at 4.1%, and ADP’s 90,000 has already set expectations a little higher. A hot number makes the December hike a certainty rather than a probability.

October 6. The next phase of tariff refunds opens, covering roughly $11.4 billion. Watch for the first legal challenges to how the money is allocated.

October 7. The minutes of the September meeting. The vote was 12-0, so the argument to look for is not whether to hike but how fast to keep going.

October 14. September CPI. This is the print that decides whether the diesel record and the Gulf refining bottleneck have made it into the headline. If they have, the December hike stops being a market expectation and becomes a Fed commitment.

October 27 and 28. The Fed meets with no projections and no expectation of a move. The interesting number is the vote, not the statement. A unanimous hold followed by a hawkish December is the base case; a dissent in favor of hiking now would be a signal that the committee thinks it is behind.

November 3 and 4. The midterm election, then the refunding the next morning. Forecasters have Democrats strongly favored in the House and slightly favored in the Senate, and a contested or slow-counted result during a war adds a risk premium rather than removing one (Cornell Chronicle; Silver Bulletin). The refunding is the single most informative scheduled event in this window, because it answers the buyback question in public.

November 26. Thanksgiving. This is when the diesel bill stops being a statistic. The federal heating assistance program faces a distribution deadline on November 1, and the households that depend on it will be buying oil into a market where the physical barrel is $121 (Congress.gov).

December 4 through December 11. The cluster. November employment on December 4. The Fed on December 8 and 9, with a fresh dot plot, where a quarter-point hike is the base case. The November inflation report on December 10. And the continuing resolution, which funds the government through December 11, expiring the following day (Congress.gov). A rate decision, an inflation print, and a funding cliff inside seventy-two hours, five weeks after an election that will decide who has to negotiate the last of the three.

That is the base case. It lands the ten-year somewhere between 5.2% and 5.8%, the thirty-year between 5.5% and 6.1%, and mortgages at 7.25% or above, with the 10s30s still pinned in the thirties and the 2s10s as the number that tells you which way the committee is leaning.

Two branches sit on either side of it, and this month they have changed shape.

The escalation branch. A full closure of the strait, or strikes that draw Iran’s neighbors in directly, removes a fifth of the world’s oil in a stroke. That is the 1979 scenario, and 1979 is the precedent that matters because it is not about the oil shock itself but about what a central bank does when a supply shock will not leave. The tightening began in 1979 and produced its damage slowly: a federal funds rate pushed toward 20% and a deep recession before inflation broke (Federal Reserve History). Bonds sell off hard at the long end no matter what the Fed does, and the question stops being whether yields reach 6% and becomes whether the Fed is willing to do what it takes to make 6% temporary.

The paradox branch. For bonds the paradox has always been the flight to quality: a genuine risk event drags capital into Treasuries and yields fall even as the news gets worse. That reflex, formed in 1990, is what a wartime bond market is supposed to do, and it has not fired once in eight months. The reason it has not fired is the same reason the strait reopening did not help. Flight to quality works when the shock is a financial event that the central bank can answer by cutting. This shock is a price event, and cutting into it is the one thing the Fed has said it will not do. So the paradox branch for this window is not peace and it is not war. It is a demand shock: a credit event, a labor market break, or a recession that arrives on its own and gives the Fed a reason to move that has nothing to do with oil. If that comes, the long end rallies hard and fast, precisely because it has spent eight months being told the inflation force wins. That is the single largest risk to everything above, and it is the one event this series has not been able to date.

What a Careful Investor Does

None of this is a prediction that yields rise in a straight line. It is a statement of asymmetry: with the long end at a twenty-four-year high and the inflation impulse no longer tied to a chokepoint that can reopen, the risks to long-term yields are skewed upward, and the price of being wrong about duration is much higher than the reward for being right. The playbook is about structure, not direction.

  • Keep duration short at the front end. Bills and short notes let you reinvest as the curve moves instead of locking a long-end loss into your statement. The two-year at 4.88% is a defensible home for money you will need within a few years. The thirty-year at 5.64% pays you fourteen basis points more than the ten-year for twenty extra years of inflation and issuance risk, which is not a bargain at any price (U.S. Treasury).
  • Own inflation directly. TIPS do work the nominal coupon cannot, and the case for them is stronger this month than last, because the inflation impulse has stopped being something a ceasefire would fix. The University of Michigan’s respondents expect 4.6% inflation over the next year while the ten-year offers 5.29% (University of Michigan).
  • Do not treat the compressed 10s30s as a signal, and do not treat it as a bargain. At 35 basis points the spread is neither warning you nor rewarding you. Watch the 2s10s instead, because that is where the market’s view of the Fed’s resolve is now written.
  • Ladder, and do not time. Nobody knows whether December brings a hike or a broken labor market. A ladder converts that ignorance into a schedule, and it is the only structure that survives both branches intact.
  • Keep liquidity for the paradox branch. The flight-to-quality rally, if it comes, will look identical to a peace rally in the first hour and be the opposite trade by the second week. Cash is what lets you tell them apart before you commit.

The Long Way of Saying It

The bond market is not a casino and it is not a democracy. It is an ongoing referendum on whether a government’s promises are worth holding, and over the next ninety days that referendum has four questions on the ballot: a central bank that has told the market to expect one more hike; a Treasury department that advertised a larger bid for its own long bonds and then declined to pay it; a strait that is reopening without making anything cheaper; and an election that will decide how much of this anyone in Washington is willing to fix.

There is a fifth question underneath the other four this month, and it is the one the last installment got wrong. I assumed the risk to the bearish case was peace, because peace would take the oil premium out and let the long end rally. The market has spent four weeks demonstrating that the premium is not in the strait. It is in the refineries that are still damaged, the insurers who still will not write the route cheaply, and the reserve that is at a forty-four-year low. A war can end in a week and none of those three fixes itself in a quarter.

That is why the level call arrived seven weeks early. The market was not waiting for news. It was repricing the arithmetic, and the arithmetic did not need the war to continue in order to stay bad. The ten-year has risen for seven consecutive months, the first streak since 2011, and it has done so through a Fed hike, a soft inflation print, and the collapse of peace talks. Storms do not send invitations, but they do leave clues, and the clue this month is that the market rose on all three.

Ninety days is not a long time. It is long enough to have your defenses built before the lightning strikes, and long enough to remember that the opportunity hidden in uncertainty belongs to the ones who kept their powder dry.

PRH | huffmanwrites.org | © Philip Huffman

Sources