The candidate pool has two layers. The forever tiers, Custodia and Vallum, draw from the large-and-mid-cap universe of every major developed market: each market's main index, or its large-caps where no index fits, plus anything large that has listed since the last rebuild. Specula casts wider, sweeping the full common-stock tape of four venues, the whole US market plus Australia, London and Taiwan, and keeping the frontier-theme names, so a sub-index micro-cap can still surface. All of it is deduped to one native listing per company, restricted to broker-tradeable stocks, and excludes mainland-China shares by policy; a market with no direct access is reached through its US or London depositary receipts. The table below lists each market, its index, the live count, and the last-fetch date.
SelectionEvery candidate must clear three assessments, applied in order: Durability, then Growth, then Confidence. A failure at any stage eliminates the company. Those that pass are ranked into Custodia by their Durability score and into Vallum by lifetime CAGR. Each tier's full Ranking Criteria are set out on its own page.
DurabilityA 0-to-100 rating of how firmly a company can hold its market leadership into an unforeseeable future, further out than any financial forecast can credibly reach. It measures the strength and defensibility of the moat, not the cyclicality of earnings. The rating is set by a multi-pass AI panel, calibrated against a fixed set of anchor companies and reconciled into a single figure. It is the gate a company must clear to reach the forever tiers, and the measure that ranks Custodia.
GrowthA test that the underlying business is still compounding, not merely the share price. Over up to ten years of revenue and operating cash flow, at least one of the two must show a rising trend once outlier years are set aside. It excludes companies whose share price has run ahead of a business that has stopped growing.
ConfidenceThe market's enduring conviction in the company, read from its long-run price. A price that keeps setting higher highs is investors repeatedly choosing to own more of the business at richer valuations, the clearest outside evidence that its lead is real and still widening. Three things must hold: prices set progressively higher peaks, never fall more than 45% from peak to trough outside a broad market crash, and do not drift behind industry peers while those peers climb.
Seedling ScoreThe score that ranks Specula. Each seedling is rated on the seven qualitative dimensions of the Specula Ranking Criteria, Asymmetry, Trend Score, Product Proof, Leadership Quality, Unit Economics, Structural Validity and Survivability, each scored 1 to 10, then combined with its one quantitative input, Revenue Growth, into a single Seedling Score.
Track RecordDurability is a claim about lasting dominance, and a company without a long enough history cannot credibly earn it, so a young company is held back from Custodia however strong it looks today. Vallum asks less, since a few years of price history are enough to establish a real growth rate and mark a Champion. As a rough guide this means roughly five years of trading for Custodia and about two for Vallum, below which a short run of returns reflects an IPO surge rather than a durable rate.
DeduplicationEach company appears once. Where the same business trades in more than one form, only the most representative listing is kept:• multiple share classes, such as Class A and Class B, collapse to one line
• common and preferred stock resolve to the common
• a holding company gives way to its operating company
• a depositary receipt gives way to the native listing, so 2330.TW is held rather than TSM
• a company cross-listed on several exchanges is held only at its primary listing
Rebuild ScheduleSemi-annual, January 1 and July 1: a full ground-up re-qualification that re-fetches the universe, re-runs every assessment, and rebuilds all tiers. Nothing has tenure.
IPO IntakeBuilding the universe from index membership alone would miss companies that never join a tracked index, such as SpaceX, and lag those that eventually do. Two backstops close the gap: a maintained watchlist of large private companies, among them OpenAI, Anthropic, Stripe and Anduril, each checked at every rebuild for a listing or other liquidity event; and a web search for any other large listing since the last rebuild. A newly public company enters the universe at once, then must build the track record it needs before it can reach the forever tiers.
Known GapsNo pool is exhaustive. The forever tiers draw on the major indexes, the large-cap sweep and the IPO backstops, so a small company outside all three can arrive late. Specula reaches deeper, down to the full common-stock listings of the US, Australia, London and Taiwan, but a small company on an exchange it does not yet sweep, such as Canada or much of continental Europe, is caught only when the frontier scan or the IPO search surfaces it. And every figure is fixed at the rebuild, so nothing here reflects price moves since.