Lab

Is the hyperscaler cohort turning? The Mag complex spent ~19 months sideways on capex fear while every benchmark ran. Daily board on the Six plus the ORCL torque row: hyperscalers-vs-consumer-AI regime line, the MAGS price map (69.50 shelf · 62.56 pivot · ~56 trendline), per-name benchmark relatives. TSLA excluded by methodology.

Who funds the AI build-out, and at what price. Six funding models from balance-sheet to vendor-financed, scored from filings — against a credit environment read in percentiles, because spreads are tight and only velocity carries information.

Where is AI's value moving? The next phase of the AI trade begins when running agents becomes cheaper than the work they replace — and value hands off, stage by stage, from the companies building the AI factory (Rubin) to the companies operating it (Agentic Infrastructure) to the companies using it (Winners). Nine ratios of our own sub-indices measure each handoff — staged in the order a regime change would propagate, each normalized to 1.00 at the 2026-06-30 reconstitution, with the thesis direction stated on every card.

Beta Instability: the chart can hold while participation breaks underneath — returns split into up-days and down-days (upside vs downside beta, asymmetry, beta-z, a 21/63/126/252-day deterioration scan) across 343 names vs SPY and Nasdaq, reproducing the "Troops Leaving Before the Generals" XSD call from the raw data. Breadth-of-Breadth: 16 structural participation proxies — equal-weight, ex-Mag-7, small-cap, credit, cyclical — distilled into one diffusion score, regime and a hidden-weakness alert.

Most screens show the sector's price; this shows what its members are doing underneath — is the move broad, or carried by a few mega-caps? For 25 ETFs & index cuts (S&P 500 sectors, global tech, Nasdaq-100 tech / ex-tech / top-30, and the software · semis · AI · cyber · quantum thematics): the median constituent vs the index, advancers vs decliners, how many names beat the index, and a cumulative advance/decline line that tends to turn before price.

Methodology hub for every Closelook pattern — PEAD (post-earnings drift), Pattern 03 (NVIDIA Strategic Authority, XIRR +109%), Sector Relative Strength, Support Confluence, Regime Signature. Each pattern documented, scored, and linked to its live signal.

Where agentic AI demand emerges, and which value-capture layer monetises it first. 8 pulses across 6 layers (App Builders · Tollbooths · Infrastructure · Endpoints · Security · Industrial), each one a multi-basket equity-gap signal. EIP / EWP / CCP daily-to-hourly, EAP / TBP / ASP weekly, CEP / IAP monthly. Read it as a stack — the layer that moves first tends to be the layer where unpriced cohorts still exist.