Daily trends: Bonds still set the agenda, but overnight the AI buildout got a delivery scare and the Iran oil premium started pricing a ceasefire float. 1. Discount rates (louder). The 30-year poked about 5.44%, highest since 2004, while Fed speakers still leave room for another hike this year. Same machine as yesterday: higher long rates reprice equity duration before they cleanly reprice the funds path. 2. Housing rates (louder). Mortgage News Daily’s 30-year fixed jumped to 7.45% Thursday, up 19 bp from 7.26%. Freddie’s weekly print is softer at 7.03%, so do not mix the two. $ITB stays on the sideline until the rent valve actually opens. 3. Diesel export float (continuing). Advisers are still analyzing a short-term ban, Europe diesel is racing the US on export-ban risk, and Wright is still arguing an outright ban eventually cuts runs. Not law yet. First-order names stay $VLO $MPC $PSX. 4. AI infra delivery (NEW). Oracle sent a force majeure notice on the Blue Owl / Project Jupiter New Mexico campus, trying to delay payments if power slips and 2028 slips. Markets had been pricing gigawatts as if they convert cleanly into contracted tokens. $ORCL $OWL are the first clips. 5. Iran / Hormuz (flipped). Tape went from widen-the-war oil surge to a 7-day ceasefire offer plus talk of a phased Hormuz reopen. Friday Asia oil already softened on that float. Until cargoes actually move, treat it as a premium float and not a peace print.

US 30-year just tagged about 5.48%, highest since 2004. The 10-year pushed through about 5.20%. 30-year mortgage rates are back around 7%. Warsh’s first hike was a front-end print. The long end is the financing vote for everything else. Equity multiples, hyperscaler paper, and the same AI buildout that still says we are not pausing all discount off a clearing rate markets have not lived with in two decades. Growth can keep equities bid while the risk-free rate climbs. Watch whether the 10-year holds above 5.20% into next week’s auctions. If it does, 5% stops being a headline and becomes the price of duration on previous-cycle PE and private credit.

Daily trends: Markets are letting the bond selloff set the agenda again. A few overnight stories got rewritten while equities took the hit. 1. Discount rates (louder). The 10-year settled about 5.11% Wednesday, highest close since 2007 after an intraday poke near 5.14%. That is not a vibes print. Higher risk-free rates compress equity multiples, refinance PE and private credit, and reprice every duration trade at once. 2. Housing rates (louder). MBA’s 30-year fixed jumped to 7.12% for the week ending Sep 18, highest since May 2024, and ARM share rose to about 9.8% as borrowers chase the teaser. Still keeps me on the sideline for the broad builder tape ($ITB) until the rent valve actually opens. Rates this high keep single-name balance sheet clearance more interesting than the ETF. 3. Diesel export ban (flipped). Energy Secretary Wright said a blanket ban will not work and pointed at voluntary flow tweaks instead. The White House denied a reported 90-day ban. The float still repriced the export arb earlier this week ($VLO $MPC $PSX), but the hard-ban path just got a hard no from the energy desk. Diesel pain is still loud. The EO is less so. 4. US-China trade clock (new). Bessent says the Busan truce extends to Jan 10, about two months, not the longer pause a lot of desks wanted. That is inventory planning time, not a durable rare-earth reset. Watch deliverables, not the tarmac photos. 5. Private credit wrappers (continuing, louder). Another big retail private-income fund capped withdrawals at the usual 5% quarterly gate after requests cleared double digits again. Gates prove the semi-liquid wrapper cannot honor simultaneous exits from illiquid loans. They do not by themselves prove credit losses yet. Pair that with 5% Treasuries and the refinance wall gets uglier. What cuts this read: yields rolling over, or a real US diesel supply add that is not an export kill switch. Until then markets are pricing higher for longer through the equity multiple.

Berkshire just crossed the 10% owner line on $LEN. Form 3 flagged Sep 17; Form 4 shows another ~2.7m Class A shares bought Sep 17–21 (~$212m), taking the Class A pile to ~23.7m. Stock popped on the filing. Markets treated homebuilders like a rates casualty. Buffett treated them like a balance sheet clearance sale. That's the machine and not a soft landing call. He's stacking voting equity in a beaten up builder while the Fed is still hiking into an energy shock. I already said I'd keep $ITB on the sideline until cuts actually move the rent valve. This doesn't flip that. It just says the cheapest way to underwrite the cycle might be one name Berkshire already underwrote and not the ETF

The $META Muse trade isn't "AI is back." it's that switching costs just got arbed away. For years telecom, auto insurance, and OTAs printed margin on consumer inertia, that people don't shop plans because sitting on hold for an hour sucks. Meta's Muse is basically a free switching bot. Goldman even runs a "consumer inertia" basket ($T $TMUS $ALL $PGR $BKNG $EXPE) and it got clipped about 2.6% in a day when the room priced that in. $ALL ate roughly 5.5%. I'd say that these companies are going through their software moment where they're going to need to prove that they can still continue to grow unaffected by agentic AI. Watch retention and shop rates next quarter on $ALL $PGR. if those gap open, this stops being a trading-desk meme and shows up in the P&L. The funny part: the same agentic tape is bidding the rails while repricing the lazy-customer moat.

Trump is floating a diesel export ban to fight $6.53 pump prices while Bessent is examining full vs partial. not law yet but having them both even dare float the idea of it is gonna cause panic in the diesel markets Gulf Coast refiners clear surplus distillate on the water, approx 1.6 mb/d lately. trap those barrels and the export arb dies, diesel cracks compress, $VLO $MPC $PSX get clipped first. US retail can ease in surplus PADDs. Europe and LatAm lose a supplier and world diesel goes the other way. if runs get cut, gasoline and jet tighten too. ironically the ban might even see US producing less diesel at a time of global shortage because the US coys are less incentivised to produce diesel if there’s a local surplus that they can’t sell. higher prices in the diesel market is an automatic incentive to increase production. so the CPI pipe doesn't magically heal. you just export the shortage to allies.

everyone keeps waiting for the housing crash like it's 2008 again. 30-year mortgages are back above 7%. that doesn't wreck the household balance sheet the way it did last time. ARM share of apps is around 8% now vs a 35% ish peak in 2005. notably, there hasn’t been any spike in ARM applications either which means there are very few active ARMs. what high mortgage does do is shove would-be buyers into apartments, so rents keep climbing while entry-level sales go soft. the release valve is the renter, not a forced liquidation wave. "wait for the crash" is the wrong trade in a rising-rate tape with locked-in owners. headwind remains for home builders $ITB and ill remain on the sideline for this till we’re back to a rate cut regime

Trump admin is in the phase of finding out that when you cut aids and the flow of money to 'allies', their ability to influence the actions of the same 'allies' drops

Considering it learnt from the best financial professionals, I’d say it’s functioning as expected

the scariest part about this is, Trump's actions convinced iran that it will never be safe unless it has nukes. maybe there was a time when they were genuinely willing to give up nukes in exchange for being accepted in the international community, but keeping them outcast and with the recent attacks only caused whatever slim chances to vanish

BofA circled two bottoms. Only one of them was a bull market. Dec 1959 (0%) and Sep 1981 (3%) both marked the end of a lost decade for long Treasuries. Both then turned. They did not turn for the same reason and they did not pay the same. What flipped Dec 1959: The 1949–59 window was the unwind of the wartime 2.5% yield peg after the 1951 Treasury-Fed Accord. Long yields ground from 2.3% to 4.7%. Coupons were too thin to offset the price losses. The last shove was the 1958 Treasury crash into the Oct 59 “Magic 5s.” The turn was a recession, not a regime change. Apr 1960–Feb 1961. Fed cut the discount rate 4% → 3%. Bills 4.57% → 2.3%. The 10-year fell from 4.72% to 3.80% in eight months. Long bonds returned +14% in 1960. New buyers locked in 4.5%, not 2.5%. That was the whole bounce. Inflation was already 1-2%. Yields only needed to stop rising. They did for a few years, then climbed again toward 7% as Vietnam and the Great Society heated up. Rolling 10-year returns recovered to a 2-4% band and sat there for 15 years. Cyclical pause. Not a bull market. What flipped Sep 1981: Different animal. Nixon shock, two oil shocks, a Fed that kept easing too early. CPI 14.8%. 10-year from 6% in 1971 to 15.84% on 30 Sep 1981. 30-year 15.25%. Funds at 19-20%. Volcker broke inflation for real. Two recessions. Unemployment 10.8%. CPI from 13-15% to under 5% by late 82. Then yields fell for 40 years. 15% starting coupon + falling rates is a machine. Long Treasuries: +33% in 1982, +26% in 1985, +24% in 1986. The rolling 10-year line went from 3% to 16% by the early 90s. That spike on the right of the chart is the Volcker bull, not mean reversion. The parallel that actually exists: Both turns needed three things at once. 1. The rate-rise that caused the losses exhausted itself. 2. A recession forced the Fed to ease. 3. The new buyer got a higher coupon than the cohort that just got run over. Starting yield is the dominant predictor of the next decade’s nominal bond return. Window math does the rest, bad years roll off, better years roll on. The parallel that does not exist: 1981’s starting yield. Today’s 10-year is ~5.0%. That is December 1959, not September 1981. The last 10 years were worse than 1959 because the starting yield in 2016–20 was 1-2%, then 0.5%. No coupon cushion. Holders who bought the pandemic low got confiscated. That part of the original post is right, and it is why sovereigns look at gold. Going forward, you get a real coupon again. You do not get 15% locked in while inflation collapses. Inflation has cooled from the 2022 peak. It has not been crushed into a multi-decade 2-3% box the way 1983-2020 was. Fiscal supply is heavier than either prior episode. Official foreign demand is lighter. So the base case is the 1959 shape: the rolling line can turn up from here if yields stop rising, because 2016-22 starts rolling out of the window and a 5% coupon finally works in your favor. That is a better decade than the last one. It is not a 40-year bull unless inflation is broken the way Volcker broke it and deficits stop fighting the bid. BofA’s title is a statement about the past. The next print depends on whether 5% is the high or just a rest stop. $TLT

Looks like Cursor created an Ultra-lite category for supergrok heavy users that got grandfathered into the promo when they signed up for supergrok heavy. 😟😟

$GNRC is ripping like Amazon just booked an $8B AI power backlog. But within the 8-K, Generac issued Amazon a warrant for up to 1.69m shares at $201, and only 308k vested day one. The rest only vests in tranches as Amazon actually pays Generac for DC backup gens, up to that $8B ceiling (aggregate gross payments, net of offsets that's a vesting meter, not a booked backlog). Near-term deliveries they're calling expected are $2.4B in 2027 and 2028. That's 30% of the meter, not 30% of Generac revenue. Full warrant is only 2.9% of the 59.0m shares outstanding and the unvested pile isn't live float today (exercise window to 16 Sep 2033). Cause this is less "Amazon just signed $8B" and more "here's a revenue progress bar tied to offtake."

The new strategy for US in the battle of the petrodollar isn't about securing supplies for themselves. It's to destroy all of their competitors so that they're the only remaining supplier