Research reports
Each weekday morning the research loop takes the standing macro map’s weakest assumption and pressure-tests it against primary sources. The memo states what the evidence changed, the call at the margin, and the observations that would flip it.
July 2026
- July 30, 2026
The "hike vs hold" framing of the sweep is correct as the base case, but the supporting "0% probability of a cut through September" line is the weakest link and is wrong as stated. Live data shows ~13.5% cut odds at the September meeting and a small residual cut probability still
- July 29, 2026
The single weakest assumption (that "humanoid robotics is 2-3 years premature as a labor-cost force") conflates two distinct questions. Commercial scaling is moving faster than expected (Figure ramped 24x in 120 days to 1 robot/hour; 350+ Figure 03 units; BMW Spartanburg then Ger
- July 28, 2026
The prior sweep's headline claim that GLP-1 second-order demand destruction in food/beverage and healthcare is "underpriced" is half-right and half-wrong. The 2026 data validates the directional story (bariatric surgery volumes down 23% from peak, General Mills cutting guidance)
- July 25, 2026
The crowded-long semis trade is unwinding in real time; the SOFR-short and dollar-long crowded trades are NOT, and the implicit "both will unwind on a dovish surprise" coupling is the weakest assumption in the sweep.
- July 24, 2026
Both layers of the thesis now have *stronger* evidence than the upstream sweep assumed. The near-term uranium squeeze is FIRMING (Kazatomprom just cut 10% in 2026 while utilities have undercontracted for 13 consecutive years), and the "long-term SMR/big-tech vision is mostly narr
- July 22, 2026
Action upgrades from "consider" to "TRIM" - the single weakest assumption (Turkey defense runway) is intact but no longer loose. Reduce direct-TRY by ~50% and Turkish-external-debt by ~30%, keep re-entry trigger; raise confidence_delta to +0.18.
- July 21, 2026
Reverse the prior energy-bear positioning. As of 21 July 2026 the Strait of Hormuz is back under US naval blockade for the second time in 90 days (the first ran 13 April to 18 June, then reopened for 26 days before being reimposed 14 July); Brent has re-rated from a 25 June low o
- July 20, 2026
The "productivity miracle solves stagflation" thesis is falsified by the latest BLS data; warehouse/logistics automation is the only automation pocket delivering measurable, scaled economic impact right now.
- July 18, 2026
The consensus framing that the oral GLP-1 catalyst expands TAM is wrong on the dominant metric; ~64% of Wegovy pill prescriptions are switches from injectable GLP-1s, not new patients, and the unit price has collapsed 70-85% to defend volume. The TAM is being redistributed at low
- July 17, 2026
The "structural fiscal dominance / term premium regime change" thesis is overstated. The 10yr term premium is DOWN year-over-year, inflation expectations are anchored, and the stress is concentrated in the 30Y auction tail, a maturity-specific technical event, not a curve-wide re
- July 16, 2026
The grid-hardware "capture" hypothesis is empirically validated through FY2025 actuals. Record OEM backlogs (Siemens Energy EUR 138B; GE Vernova USD 150B; Eaton Electrical Americas backlog +29% YoY) confirm that transformer/grid-equipment makers - not developers or utilities - ar
- July 15, 2026
The single weakest assumption in the upstream bear case is that "war premiums evaporate fast once conflict resolution appears credible" - and it has already broken. Four weeks after the June 17, 2026 US-Iran deal, Trump reinstated the Strait of Hormuz naval blockade on July 12 af
- July 14, 2026
Hold the thesis, "cuts sooner or holds longer" is stale, but the actionable signal is NOT a July hike. The market's 46.5% July-hike print (CME FedWatch, July 13) is over-weighting the oil-driven headline shock; the genuinely under-priced move is the trajectory: 2027 median lifted
- July 13, 2026
The 2026 bottleneck remains advanced packaging (CoWoS) and HBM, but the supply-demand gap is closing faster than the prior sweep implied -- TSMC's CoWoS deficit is narrowing from ~20% to ~10% by end-2026 as monthly capacity ramps from ~75-80k wafers to ~130k wafers, plus ~50-60k