How is each tracked watchlist performing?
Composites are hand-built baskets — our own synthetic ETFs with chosen members and weights, tracked against a benchmark.
6 baskets, ranked by 30-day return. Add one by dropping a JSON definition in src/content/composites/.
Equal-weight composite of the proven Wave-1 optical names: the lasers, transceivers, systems, and interconnect layers of the AI photonics build-out that NVIDIA anchored with direct investment (LITE, COHR) — plus the module and retimer layers riding the same capex (FN, CIEN, ALAB, CRDO).
Weighted composite from the memory-supercycle lane (validated 2026-06-29): HBM cannibalizes DRAM, suppliers capture Mag7 capex. Overweights the DRAM/NAND primaries (MU, SNDK), carries Korea exposure via EWY, and keeps a tail in the HDD pair (STX, WDC).
Equal-weight composite of the durable, profitable tier of the robotics build-out — semiconductor test (TER), surgical robotics (ISRG), machine vision (CGNX), factory automation (ROK), and advanced-packaging equipment (KLIC). Deliberately excludes the pre-revenue humanoid spec tier.
Equal-weight composite of the tools/CRO tier of the biotech capital-cycle thesis: quality laggards that monetize a reopened biotech funding window without single-trial risk — instruments (TMO, DHR), contract research (MEDP), bioprocessing (RGEN), and computational chemistry (SDGR).
Equal-weight composite from the electricity lane: AI datacenter power demand played across contracted generation (CEG, VST, TLN), grid and thermal equipment (GEV, VRT, ETN), and the superalloy metallurgy upstream of the turbines (HWM, ATI).
Equal-weight composite from the nuclear-fuel-cycle lane: AI-driven nuclear demand, played upstream (enrichment + uranium + contracted utilities), not pre-revenue SMR developers.