The method, in plain language.

Emerging — one or two specialists have started building. Thin evidence, early.

Early Conviction — several specialists with demonstrated edge in this niche have converged. The crowd hasn't noticed.

Consensus — participation is broadening. Well positioned; the window is narrowing, not closed.

Crowded — ownership is wide, marginal buyers are late arrivals.

Fading — the specialists who built the trade are reducing while the crowd holds.

Stages describe the evidence, not your decision.

A name crosses into Crowded when the share of its institutional ownership held by non-specialist, trend-following money crosses our threshold while specialist adds have stopped — measured every weekly read, in lockstep across the whole book, using the same rule for every stock. No discretion per name.

All 2,825 disclosing funds are graded A to F on where their money sits today: overlap with converged specialist buys (conviction), weight already in Crowded or Fading names (late), and active share. Rank compares each fund with its own peer group (same segment), so a mid segment fund at 50% conviction can outrank a large segment fund at 70%. Funds outside US-equity coverage — bond, short-duration, money-market — are not graded, because our edge is absent there.

We grade managers not on average returns but on where each one has demonstrated edge: how their disclosed buys performed inside their niche, against the market, over the tracked window. The best are specialists you've never heard of. Fame and AUM carry no weight.

A receipt opens when a Print fires and closes when the opposite Print fires on the same name — find receipts run from conviction to the first warning, warning receipts from the warning to the exit of the episode. The outcome is the price change between the two timestamps. Every receipt is kept: losses included, nothing edited after the fact. Aggregate base rates shown across the product are computed from every receipt in the archive.

2026-06 — Verdict vocabulary simplified to Early · Well positioned · Late · No conviction. Warning episodes replace repeat warnings. Tape reads separated from stage prints.

2026-05 — Fund grading moved to Kensys full-book validation. Crowding flag moved to strict lockstep definition.

2023-12 — Tracked live begins. Every read timestamped since.