Macro worldview (4.0.2 - Thursday September 10 patch - the long end broke through a tripled Treasury buyback with the 30Y at 5.307% and the 10Y at 4.85%, August payrolls rebounded to +162K with +55K of upward revisions, August PPI ran 5.4% YoY on a 24.1% diesel surge while core softened to +0.2%, the Gulf war became a tanker war with Brent above $101, and gold's structural leg refreshed on a record $18B August ETF inflow. No thesis added, retired or invalidated; no invalidation condition fires)

4.0.2 refresh, taken Thursday September 10 ~12:01 PM ET - the window runs from the pre-payrolls session of the September 3 snapshot (4.0.1) through August payrolls on September 4, the tanker-war escalation of September 5 to 9, AMD's Citi TMT guide on September 8, the tripled Treasury buyback on September 9 and into this morning's August PPI print, six days ahead of the FOMC. This is a patch. No thesis is added, retired or invalidated, no invalidation condition fires, and the 4.0.1 framing - a hawkish regime, a restored energy premium and a broken long end - is not rewritten but tested from three directions at once. Net moves: Persistent energy premium +0.05 to 0.85 (band 0.07 to 0.06), Gold structural debasement bid +0.03 to 0.89 (band 0.06 to 0.05), Hawkish Warsh regime and higher-for-longer +0.02 to 0.89, AI capex sustained but with China decoupling tail risk +0.02 to 0.92 (band 0.04 to 0.05), Iran war rearmament cycle +0.01 to 0.90, Equity melt-up versus building recession risk -0.01 to 0.77 (band 0.06 to 0.07), Stagflation risk and Fed independence stress held at 0.90 with its band widened to 0.05.

The long end refused a direct intervention, which is the single most important fact in the window. Hawkish Warsh regime and higher-for-longer steps 0.87 to 0.89. On September 9 Treasury announced it would buy back up to $6B of longer-dated debt, triple the $2B initially communicated, an explicit Bessent effort to put a lid on borrowing costs - and it failed in the session it was announced. The 30-year rose five basis points through the 5.3% level to 5.307% and the 10-year reached 4.85%, its highest since October 2023, with TLT down to 81.00 this morning from 81.95 on September 2. A long end that sells off into a tripled buyback is telling you the bid is not coming back at these levels, and it is doing so while September hike odds have gone 57.5% to 60.6%. The supporting data cooperated: August payrolls came in at +162,000 against a +53,000 consensus, the strongest print since March and the prior two months were revised up a combined 55,000, turning July from -23,000 to +21,000, removing the labour-softness argument for holding. The band is held rather than tightened at 0.04 for one honest reason: core PPI rose only 0.2% in August against a 0.3% forecast, the first genuinely softer inflation reading this thesis has faced, and a benign CPI tomorrow could still produce a hold on September 16.

Inflation strengthened and the growth leg moved the wrong way, so the mean holds and the band widens. Stagflation risk and Fed independence stress stays at 0.90 with its band 0.04 to 0.05 - the only band widened without a mean move. The inflation half is unambiguous: August headline PPI rose 0.4% on the month to 5.4% year-over-year, up from 4.8% in July, with over a third of the goods increase traced to diesel prices, which jumped 24.1% in a single month, and the ISM Services Prices Index hit 72.6 in August, its highest since 2022. That diesel number is the energy premium arriving in producer prices in the most direct form yet. But the stag half receded again: payrolls rebounded hard, ISM Services printed 55.4 with new orders at 60.9, a 26th month of expansion, and unemployment held at 4.1%. The two genuine counterweights are narrow - the ISM Services employment index stayed in contraction at 47.8, and average hourly earnings grew 3.1% year-over-year against a headline CPI consensus of 3.4%, leaving real wages negative. A thesis whose two legs now openly disagree deserves a wider band, not a moved mean.

The Gulf war became a tanker war and the premium extended. Persistent energy premium steps 0.80 to 0.85, the largest move in this refresh. The IRGC targeted a US Navy warship with ballistic missiles twice over two days, the United States destroyed five Iranian crude tankers in the Gulf of Oman on September 8 in retaliation, and Iran struck two American vessels and eight oil tankers and fired missiles at US targets in Jordan. Price followed decisively: Brent settled at $101.21 on September 9, its highest close since May 22, with WTI at $96.05, Brent reached $105.20 intraday this morning, and the USO oil fund rose 10.6% across the window from 141.15 to 156.14. The step is +0.05 rather than more, and the band only tightens to 0.06, because the contrary case got better too: Iran and Oman are negotiating a phased framework for a temporary shipping corridor through the strait including joint mine-clearing, and Goldman raised its December 2026 Brent forecast by $5 to $85 - still sixteen dollars BELOW spot, which is the sell-side saying this premium is a war premium that decays. The invalidation needs a durable reopening followed by WTI under $80 for thirty days; this window moved that further away while simultaneously building the machinery that could deliver it.

Gold's structural leg refreshed with September-dated evidence, which is exactly what 4.0.1 said was missing. Gold structural debasement bid recovers 0.86 to 0.89 and the band tightens 0.06 to 0.05. The prior snapshot marked this thesis down because gold broke with the debasement story on rising real yields, and explicitly noted its structural supports were Q2 facts rather than September facts. That gap is now closed: global gold ETFs took in $18B in August, the second-largest monthly inflow ever recorded, with European funds posting their biggest month on record at $7.9B and North America its third-largest at $7.7B, and the PBoC added roughly 20 tonnes in August, extending its buying streak to 22 consecutive months, the longest on record, past 2,600 tonnes. Meanwhile the price stopped falling: spot recovered to $4,399 on September 9, up 1.0%, after a three-session slide and GLD closed 403.35 with GDX at 99.47, both above their September 2 levels - and it did so while the 10-year TIPS real yield tested its 2025 high near 2.44%. Holding a bid against the very mechanism that broke it a week ago is the strongest form this evidence could take. Held below 0.90 because gold remains roughly 5% under its August 28 level.

AI capex broadened decisively, and for the first time its financing became visible as a risk. AI capex sustained but with China decoupling tail risk steps 0.90 to 0.92 while its band WIDENS 0.04 to 0.05 - a deliberate pairing. The demand evidence is the strongest this thesis has ever carried: Broadcom's fiscal Q3 AI semiconductor revenue tripled to $16.7B, up 221% year-over-year and 54% sequentially against a $15.2B estimate, and management guided FY2026 AI revenue to $58B and outlined roughly $115B for FY2027 and $230B for FY2028 - forward visibility, not just a print. Breadth arrived too: AMD told the Citi TMT conference on September 8 that data-center revenue will roughly double to about $70B in 2027 with AI GPUs alone contributing low-$40B, with MI450 and Helios already shipping, CLSA lifted its price objective from $575 to $710, and AMD rose 14% across the window from 457.06 to 521.10. Against that, two facts cut the other way and are why the band widens rather than tightens: Broadcom fell 2.7% on September 3 despite the beat, on a Q4 revenue guide of $34.8B just under the ~$35B consensus, and JPMorgan calculates the hyperscalers plus NVIDIA have issued roughly $320B of debt in 2026, equal in ten-year-equivalent terms to about 68% of new long-duration Treasury borrowing this year. The AI buildout is now a first-order participant in the long-end selloff described above - which means this thesis and Hawkish Warsh regime and higher-for-longer are no longer independent, and a capex cycle financed into a 5.3% thirty-year carries a cost vulnerability it did not carry in June.

The index held; what is underneath it did not. Equity melt-up versus building recession risk eases 0.78 to 0.77 with its band 0.06 to 0.07. The headline resilience is real and deserves stating plainly: against Brent above $100 and a 4.85% ten-year, SPY closed September 9 at 762.40, down only 0.4% from September 2, while QQQ actually ROSE 1.0% to 716.31, and the S&P 500 sat at 7,636.36. That is melt-up behaviour under a genuinely hostile tape. But the composition kept deteriorating: the Russell 2000 fell 1.9% across the window to 290.64 and another 0.8% this morning to 288.36, underperforming the S&P by a widening margin, and Washington unveiled an import ban on a range of Canadian goods on September 9, largely replacing the 50% tariffs, to take effect September 29 after Canada's retaliatory tariffs of 15 to 50% on nearly $20B of US imports took effect September 8 - a second inflationary supply shock forming alongside the energy one. Both invalidation legs were checked and neither fires: the VIX closed 16.44 on September 9 and traded 16.08 this morning, above the 15 the melt-up-completion leg requires and far below the 25 of the vol-expansion leg. The mean moves only -0.01 because the large-cap tape genuinely refused to break; the band widens because large caps and small caps are now telling different stories.

Rearmament escalated again and the equity de-rated again. Iran war rearmament cycle steps 0.89 to 0.90 with its band held. The window delivered the widest exchange of fire of the war and, newly captured here, the Navy awarded Raytheon a $22.9B contract on August 17 to accelerate Tomahawk production - contracted multi-year procurement that no diplomatic outcome unwinds quickly, and a fact prior snapshots missed. Yet the primes fell again through the escalation: LMT 531.55 to 524.46, RTX 200.78 to 197.55 and NOC 523.82 to 515.57 between September 2 and 9. This is the second consecutive refresh in which defense equities decline during an escalation week. That is not evidence against the thesis - it is precisely the backlog-versus-price gap the thesis exists to express, and it keeps widening - but two consecutive windows of it is the reason the step stays at +0.01 and the band does not tighten.

Catalyst calendar. August CPI lands tomorrow, Friday September 11, with consensus at 3.4% - the first inflation print carrying the reversed energy base in full, and after this morning's split PPI it is the decisive input for whether Hawkish Warsh regime and higher-for-longer gets its hike. The September 16 FOMC follows six days later at roughly 60% priced: a hike confirms the regime, a hold with hawkish language leaves it, and a dovish hold remains the first real contrary evidence this thesis would face. The Fed is in blackout until then. On energy, whether the Iran-Oman corridor framework produces anything concrete governs Persistent energy premium, and it is now the only live path toward that thesis's invalidation. The September 29 effective date of the US import ban on Canadian goods is the next step in a trade shock that currently sits inside Stagflation risk and Fed independence stress rather than in a thesis of its own - if it broadens, that judgement should be revisited.