Majors synthesis

Report date 2026-08-05 · window 2026-08-02 to 2026-08-05 (4 days, 661 document rows, 15 publishers) · book at 2026-08-04, 12,868,169 USD invested capital · all weights and percentages on the invested_capital basis
The corpus stopped disagreeing with itself on 08-04 and has not resumed. Eleven of sixteen cells have never printed a negative reading in the window; the five that did are all in Minerals and Power Generation, and all of them turned before 08-04. Nothing in four days of corpus reads against the direction of this book at the grid level — which is a fact about the corpus, not a confirmation of the book.

The grid — four majors, four horizons

1m3m6m6m+
AI+0.305 docs · contested+0.00+0.02+0.32+0.30+0.426 docs+0.00+0.07+0.45+0.42+0.405 docs+0.20+0.39+0.44+0.40+0.503 docs+0.33+0.31+0.50+0.50
Minerals+0.173 docs · contested-0.17-0.15+0.17+0.17+0.173 docs · contested+0.00-0.04+0.50+0.17+0.502 docs+0.50+0.30+0.50+0.50+0.001 doc·+0.36+0.50+0.00
Power Generation+0.3211 docs · contested-0.50+0.03+0.31+0.32+0.2311 docs · contested+0.00-0.08+0.42+0.23+0.399 docs · contested-0.50+0.28+0.25+0.39+0.448 docs+0.50+0.36+0.50+0.44
Energy Infrastructure+0.173 docs·+0.00·+0.17+0.002 docs · contested·+0.10·+0.00+0.333 docs · contested·+0.50·+0.33+0.673 docs·+0.40·+0.67

Each cell shows the latest reading, the documents under it, and the path across the window (oldest first). A grey dot is a day the cell was quiet — no document addressed it, which is not the same as a neutral score. An amber cell rests on two documents or fewer.

What moved

Where the corpus reads against this book

Forty per cent of AI document rows read against this book, from eight separate houses, under one rule: picks-and-shovels. The mandate expresses AI through memory, power and the commodity inputs, not the hyperscalers, so a bull call on hyperscaler outperformance is a call against the layer this book owns. Minerals and Energy Infrastructure produced zero inverted rows across 661; Power Generation produced eight, all from one house. The disagreement in this corpus is concentrated almost entirely in the sleeve holding 11.3 per cent of the fund, and it is one-directional.

MajorWeightRows WithAgainstAgainst % Houses reading against
AI11.3%31218612640.4%Academy Securities, Bear Traps, Deutsche Bank, Goldman Sachs, JP Morgan, TS Lombard, The Market Ear, Zero Hedge
Power Generation4.6%21821083.7%Goldman Sachs
Minerals63.4%959500.0%
Energy Infrastructure20.7%363600.0%

What the fund holds against what the research names

Three of thirty held names were written about by name across four days and 661 document rows: GOOGL in 134 rows across three days and four houses, CCJ in 49 rows on one day, AMD in 8 rows on one day. GLW and MU appear only inside emerging subjects, never in a scored document. The remaining twenty-five names, including thirteen of the fourteen Minerals positions carrying 63.4 per cent of invested capital, were not named once. The research the fund is buying reaches roughly a sixth of its book by name; the rest is covered by sleeve mapping, which is this repo's table asserting a subject bears on a major, not a publisher writing about a position.

Held nameRows naming it DaysHouses
GOOGL1343Goldman Sachs, JP Morgan, The Market Ear, Zero Hedge
CCJ491Bear Traps, Goldman Sachs
AMD81Goldman Sachs
GLW0emerging subjects only: ai-hedge-fund-blowup, bear-steepening, fed-forward-guidance
MU0emerging subjects only: enterprise-ai-adoption, etf-flows, fed-rate-path

Held positions no document named in the window, by sleeve:

SleeveSilentWeight Market value USDNames
Minerals13 of 1463.4%8,155,152AA, AUGO, B, CENX, FCX, FNV, HL, LUN, NEM, RIO, SBSW, SCCO, WPM
Power Generation2 of 24.6%593,400EQT, RIG
Energy Infrastructure10 of 1020.7%2,664,704AMRC, AZZ, ETN, GEV, MYRG, NVT, POWL, PRIM, VMI, VRT

Macro strip — out-of-major subjects that bear on the book

The strip is led by yen-intervention, seven publishers over three days, reading against Minerals through the non-USD diversification overlay. Two of the next four, fed-rate-path and fed-forward-guidance, read against Minerals through the same negative-real-rates doctrine. The concentration is the point: three of the five strongest out-of-major subjects bear on the sleeve the in-major corpus is nearly silent about, and they are the closest thing to a second opinion the window contains.

SubjectStrengthHouses DaysReads againstBasis
yen-intervention8.8873MineralsBears on the non-USD diversification overlay running alongside commodity-exporter selection.
fed-rate-path3.7442MineralsBears on the negative-real-rates doctrine, which is the case for the gold and silver allocation.
momentum-unwind3.7442AI, Energy InfrastructureBears on crowded-longs risk, the named shared risk across the AI and power-infrastructure sleeves.
china-open-source-ai-models3.1532AIBears on the AI sleeve, at the layer the book owns or one above it.
fed-forward-guidance3.1542MineralsBears on the negative-real-rates doctrine, which is the case for the gold and silver allocation.
leveraged-etf-unwind3.1532AI, Energy InfrastructureBears on crowded-longs risk, the named shared risk across the AI and power-infrastructure sleeves.
bear-steepening3.0242MineralsBears on the negative-real-rates doctrine, which is the case for the gold and silver allocation.
us-iran-de-escalation3.0242Power GenerationBears on the Hormuz crisis-premium overlay the energy sleeve carries.
fed-credibility2.9133MineralsBears on the negative-real-rates doctrine, which is the case for the gold and silver allocation.
us-iran-conflict2.8941Power GenerationBears on the Hormuz crisis-premium overlay the energy sleeve carries.

Strength is unsigned. A subject bearing on a sleeve is not a directional call on it; direction is not measured upstream and is not inferred here.

Emerging — strength only

SubjectStrengthHouses DaysSeen
yen-intervention8.887308-03–08-05promotion candidate
el-nino-risk3.744208-03–08-04
fed-rate-path3.744208-03–08-04
momentum-unwind3.744208-03–08-04
china-open-source-ai-models3.153208-03–08-04
fed-forward-guidance3.154208-03–08-04
leveraged-etf-unwind3.153208-03–08-04
bear-steepening3.024208-03–08-04
growth-to-value-rotation3.024208-03–08-04
us-iran-de-escalation3.024208-03–08-04
fed-credibility2.913308-02–08-04promotion candidate
us-iran-conflict2.894108-03–08-03

20 of 41 forwarded subjects. upstream drops subjects with fewer than 2 distinct publishers — one house filing three notes is not three houses converging.

What to watch

Minerals carries 63.4 per cent of invested capital, thirteen of its fourteen names went unmentioned for four days, its future cell rests on a single document, and every macro subject pointed at it points the same way. That combination is a coverage gap, not a signal, and it is the one thing on this page that should change what gets commissioned next.

Boundary readings applied