Oil falls and stocks finish a winning week, but long-term borrowing costs hit another two-decade high
- The thirty-year Treasury yield closed at 5.488% and the ten-year at 5.163% — the long end rose on a day oil fell 2%, so the week's bond selling no longer depends on oil alone.
- Final September consumer sentiment rose to 48.1, with year-ahead inflation expectations at 4.6% and five-year expectations up to 3.4% from 3.3% — long-run expectations moving is the survey result a Fed on a hiking path reacts to.
- CME FedWatch put the chance of an October 28 hike at about 76%, up from 71% before the open — the morning's drop in hike odds, which followed oil, was reversed by the close.
- Brent fell 2.14% to $104.32 and WTI 2.33% to $92.41 as the Dow rose 0.93% to end a three-week losing streak — stocks took the oil relief, while the long bond rose anyway.
- August durable goods orders were flat at $338.6 billion against a forecast 0.4% fall, with orders excluding transportation up 0.3% — the goods economy is holding up, which gives the Fed no growth reason to skip October.
+ 3 more sourced points ▾− show fewer ▴
Methodology note: Night report, generated after the September 25 US close. Sources cited inline.
Oil Falls 2% and Stocks Finish a Winning Week, but the Thirty-Year Closes at Another Two-Decade High
Oil fell, stocks rose, and the long bond did not follow. Brent settled 2.1% lower at $104.32 and the Dow gained 0.93%, ending a three-week losing streak. The thirty-year still closed at 5.488%, its highest since 2004, above Thursday's 5.48%. This morning's report set that as the test: a close at or above 5.48% with Brent lower would mean oil is no longer the main driver of the long end. That is what happened. The data behind it points to inflation expectations. Consumers confirmed year-ahead inflation expectations at 4.6%, and October hike odds went back up to about 76% from 71% before the open. Stocks and bonds are pricing different things. A thirty-year close under 5.40% would end that split.
- The ten-year closed at 5.163%, up less than 1 basis point, and the thirty-year at 5.488%, up about 2 basis points — the long end rose on a day oil fell 2%, so this week's selling no longer depends on oil alone.
- Final September sentiment rose to 48.1 from a preliminary 47.8, with year-ahead inflation expectations at 4.6% and five-year expectations up to 3.4% from 3.3% — the five-year figure had held at 3.3% for three months, and a rise in long-run expectations is the one survey result a Fed on a hiking path reacts to.
- August durable goods orders were flat at $338.6 billion against a forecast 0.4% fall, with orders excluding transportation up 0.3% — the goods economy is holding up, which gives the Fed no growth reason to skip October.
- CME FedWatch put the chance of an October 28 hike at about 76% — up from 71% before the open. The drop this morning, which followed oil, was reversed by the close even though oil fell further.
- Brent fell 2.14% to $104.32 and WTI 2.33% to $92.41 as Iran's seven-day Hormuz plan stayed on the record — oil fell even as Saudi Arabia intercepted six Houthi ballistic missiles aimed at Taif and Yanbu, so the market is pricing the diplomatic track ahead of the attacks.
September 25, 2026 Close
| Asset | Level | Change |
|---|---|---|
| S&P 500 | 7,743.41 | +0.51%; week +0.6% |
| Nasdaq Composite | 27,068.72 | +0.5%; Nasdaq 100 week +2.1% |
| Dow Jones | 51,828.62 | +478.64 pts, +0.93%; first weekly gain in four |
| Russell 2000 | — | close not confirmed |
| VIX | 15.67 | change not confirmed |
| 10Y UST | 5.163% | +<1 bp; week high above 5.22% |
| 30Y UST | 5.488% | +~2 bp; highest close since 2004 |
| Oct hike odds (FedWatch) | ~76% | from 71% pre-market |
| Brent (Nov) | $104.32 | -2.14% |
| WTI | $92.41 | -2.33% |
| Gold (spot) | ~$4,284 | +$10.80; down more than 2% on the week |
| DXY | ~101.15 | second weekly gain |
| BTC | ~$84,400-85,200 | intraday range; close not confirmed |
| ETH | ~$2,690-2,740 | intraday range; close not confirmed |
| Nikkei 225 | 66,364 | +1.3%, fifth straight gain |
Why it happened
- The morning read was right on oil and stocks and wrong on bonds. It expected oil's pullback to take pressure off the long end. Brent fell further than the morning's 1%, yet the thirty-year closed higher and hike odds rose.
- Inflation expectations replaced oil as the input. Year-ahead expectations at 4.6% compare with 3.4% before the Iran conflict began, and the five-year measure moved up for the first time in four months. Long bonds are priced off where inflation settles, not where Brent is this week.
- Growth data gave the Fed no reason to wait. Durable goods beat a forecast decline, following the strong flash PMI earlier in the week. With activity holding up, the October hike case rests on inflation, and the survey supplied more of it.
- The dollar and gold told the same story as the long end. The DXY held near 101.15 for a second weekly gain and gold lost more than 2% on the week despite Friday's $10.80 rise. Both are consistent with a market paying for a Fed that hikes again, not for an oil shock fading.
- Stocks priced the oil relief and the AI news. The Dow's cyclicals led as oil fell, and the Nasdaq 100's 2.1% weekly gain came from the Meta Muse and CPU trade. Neither depends on the long end in the short run, which is why equities could ignore it for another day.
Movers
Atlas Energy Solutions rose 18.2% to $12.99 after disclosing about $613.5 million of cost reimbursement agreements with an unnamed frontier AI lab for power-generation equipment, including 283 megawatts of added Caterpillar orders. It is the second supplier in two days, after Akamai, to book AI-lab spending on data-center capacity.
Akamai closed up about 3%, a fraction of the roughly 20% it was quoted premarket on the $11.6 billion Anthropic commitment. The fade says the market is valuing the contract more cautiously than the first reaction did, and the $111.33 warrant strike remains the level to watch.
Meta fell 3.3% but ended the week up about 12% after the Muse agent launch. Friday's decline is profit-taking on a large weekly move, not new information about Muse.
AMD closed at $630.12, up 188% for the year against about 22% for Nvidia. That gap, the widest of the AI boom, rests on the case that AI agents need more CPUs, and the October 1 price increase is the first test of the pricing power it implies.
What to watch
Whether the thirty-year holds above 5.44% without oil behind it. It has now closed above the 2007 peak twice, the second time on a day Brent fell 2%. A third close above 5.44% on Monday, a day with no coupon auction, would confirm the break. A close under 5.40% would say this week was auctions and one survey. The thirty-year was below 5% as recently as early July, so the move is nearly 50 basis points in under three months, and the Treasury buyback program running through November 4 has not stopped it.
August PCE on September 30 against 76% October odds. It is the first hard inflation data since the repricing, and the survey's 4.6% expectation has raised the bar for a benign print. A core PCE at or above consensus would put the ten-year's 5.22% weekly high back in play. A soft print would test whether the hike case rests on data or on expectations alone.
Whether the Hormuz plan gets a US answer by September 29. Brent is down about 2% from Thursday's $106.60 on an offer Washington has not accepted, while Houthi missiles are targeting the Saudi port at Yanbu. A US acceptance would start the seven-day clock and put Brent under $100 in reach. No answer by the 29th leaves the drop resting on an offer, not an agreement.
Next 5 Trading Days
Mon Sep 28 — No coupon auction and no major data. The cleanest read on whether the thirty-year holds above 5.44%.
Tue Sep 29 — JOLTS, Conference Board confidence and the reported Hormuz window closes. Decides whether Brent's drop has a deal under it.
Wed Sep 30 — August PCE, third-quarter-end, ADP and Micron's fiscal Q4. The inflation test and the AI-capex test on one day.
Thu Oct 1 — ISM manufacturing, Nike's fiscal Q1, and AMD's price increase takes effect. Tests the goods economy's strength and the tariff channel.
Fri Oct 2 — September payrolls after August's 162,000. A second strong print would make an October hike close to fully priced.
