ValueGrid is an illustrative visualization of global capital flows. It mixes verified filings data with modeled estimates — and tags every figure so you can tell which is which.
What this is not: not real-time market data, not audited, not investment advice. Map pins mark headquarters locations (manually curated, periodically re-verified). Market caps are point-in-time snapshots, not historical series. "Source quality" percentages express relative source strength within this dataset, not statistical confidence intervals.
| R | Reported. Backed by a primary filing we actually fetched. Company revenue tagged R comes from SEC XBRL facts (10-K / 20-F annual filings, FY2019–2025), refreshed automatically — the inspector links each one to the underlying SEC data ("Verify at SEC"). |
| E | Estimated. Modeled from disclosure plus input-output tables, tax-incidence models, or trade data. Also covers figures we could not verify against a filing (non-SEC reporters, non-USD filers) and point-in-time values like market cap. |
| I | Inferred. Third-party allocations and heuristics (bill-of-materials traces, 13-F ownership × payout, banking-relationship models). Directional only. |
The second tab ranks scarcity rents — things whose price has diverged far above the cost of new supply because supply cannot respond. Each signal carries a current price, the baseline it is priced against, the annual excess rent that gap implies, who collects it, why supply is stuck, and when relief arrives. The rank score is computed from those fields (rent multiple 30%, rent pool 28%, persistence 20%, concentration 12%, incumbent margin 10%), never hand-typed, so the ranking cannot drift from the data behind it.
Provenance is deliberately excluded from the score. A weakly-sourced signal is not quietly demoted — it ranks where the numbers put it and shows an I badge. Most price and rent-pool figures here are Estimated or Inferred: bilateral contract prices, packaging slots and CDMO rates simply are not publicly quoted. The Reverted signals archive exists so no entry reads as permanent — every rent on the board is a bet on a supply response that has not happened yet.
Estimated and Inferred figures are directional, not audited. Flow values between specific counterparties are modeled even when both endpoints report real totals. Confidence percentages express relative source quality within this dataset, not statistical confidence intervals. Do not cite modeled figures as fact.
Sources: SEC EDGAR XBRL company facts · company annual reports · UN Comtrade · BEA/OECD input-output tables · 13-F holdings · USAspending.gov · central-bank financial statements · BEA state PCE · Eurostat · China NBS · Japan Cabinet Office · World Bank HFCE (household breakdowns) · Frankfurter (ECB) · CoinGecko. Code & data: github.com/AlvanChow/xian