How the revenue mapping is built
Every figure in Strata Thesis rests on one classification: each company’s reported revenue, segment by segment, assigned to 50 themes with the reason written down. The unit is one reported revenue segment of one company. Every step below is written down, enforced in code, or both.
One classification, read three ways
The same taxonomy reads the companies, the holdings of every fund, and the 13F books, so purity means the same thing in all three places.
The Builder
Which companies qualify at any revenue threshold, every segment and why it was classified that way, and the basket the rules produce.
Competition
3,200 funds read through their latest holdings: purity on any theme, overlap with the basket, and the names no fund holds.
Demand
4,800+ RIAs and hedge funds above $100M, matched quarter to quarter: who holds the names, and whether they added or trimmed.
The five steps
- 01
Segments from filings
Each company’s reported revenue segments and their share of revenue, from the annual report: SEC 10-K and 20-F XBRL, EDINET for Japan, TWSE for Taiwan, DART for Korea, ESEF for Europe and the UK, and Chinese company reports. Subtotal, geographic and elimination lines are removed. Segments must reconcile to reported revenue within 10%, and every source document is archived. Where quarterly data passes every consistency check, segment weights are brought to a trailing twelve months; the theme labels never change in that step.
- 02
Every segment classified
Each segment is split across one to four of the 50 themes, with the share of the segment’s revenue in each, summing to 100, a confidence score for the evidence, and a one-line reason naming the product, customer or stated figure. Settings are fixed so the same input always gives the same answer, and results are cached by a hash of the input and the rules.
- 03
Boundaries enforced in code
Each theme definition says what counts and where adjacent businesses belong. The classification rules include: what a company sells, not who buys it; divested businesses get no exposure; region-named segments are classified by what they sell; captive finance and warranty businesses roll into the parent; suppliers whose segments are named after end markets are classified by what they make; the filing outranks the data vendor’s description. After classification, code applies the boundaries. Examples: equipment makers cannot hold Utilities; every non-software segment of a chip company keeps Semiconductors at 50% or more; cameras and access control are Internet of Things, not Cybersecurity; a REIT keeps Real Estate at 55% or more; nuclear fuel is Mining & Metals; stationary storage is Clean Tech. Approved reference splits sit beside the rules.
- 04
Weighting with confidence
A company’s exposure to a theme is the segment’s share of revenue times the theme’s share of the segment. Confidence is scored 90 to 100 when the filing states it directly, 70 to 89 for a strong inference, 50 to 69 when inferred from an industry label. A theme under 70 is halved. A change of a company’s primary theme against the last release needs confidence of 80 or more. A theme counts as a company’s theme when its exposure is at least half of the company’s largest theme.
- 05
Audit and release
A second, independent review reads every company and flags a wrong primary theme, a missing or extraneous theme, weights that look off, or stale data. Changes are trialed in a sandbox, diffed against the previous release and signed off. Each release is numbered (major for a taxonomy change, minor for a method change, patch for data fixes and new companies), dated and frozen with a manifest, release notes and a diff; a correction is a new release. The release and its date are printed on every page.
What the data carries
| Level | Fields |
|---|---|
| Segment line | Company, segment, share of company revenue, theme, share of the segment in that theme, confidence, reason, source (model or named code rule) |
| Company | Exposure by theme (sum of segment share times theme share), evidence basis, filing period the segments describe, classification release |
| Fund | Revenue-by-theme book from its latest holdings (each holding’s company exposure, weighted), purity on any theme or combination, signature themes at 12%+ |
| 13F book | For each constituent: holders, prevalence, median weight, matched-panel net flow quarter to quarter, firms adding and trimming |
| Advisor adoption | For each thematic ETF and quarter since mid-2023: RIAs holding it, openings and exits, retention after one to four quarters, and each adopter’s prior book read through its funds |
As of 2026-10-19: classification release v2.2 (2026-10-17), market data 2026-10-16, 13F quarter 2026-06-30, fund flows through 2026-07.