What we measure, and what we don’t.
Four countries and two dataset families. Every cell below is either populated or carries the reason it is not.
Coverage matrix
| Country | Stablecoins | Macro | Credit | RWA / Tokenization |
|---|---|---|---|---|
| Brazil | 2026-10-01 | 2026-10-01 | 2026-08-31 | 2026-09-18 |
| Argentina | 2026-10-01 | 2026-10-01 | 2026-08-31 | — not ingested |
| Colombia | 2026-10-01 | 2026-10-01 | 2026-08-31 | — not ingested |
| Mexico | 2026-10-01 | 2026-10-01 | 2026-08-31 | — not ingested |
NOT IN SCOPE — v1 covers BR · AR · MX · CO
DEX liquidityDune-sourced; embargoed until a licensing verdict is recorded
CVM filingsnot yet ingested
Stellar transfersquarantined pending re-fetch; the imported data is known-invalid
Solana transferstable created, not yet ingested
Settled value by token, weekly
PROVISIONAL · BR AR CO MX
20,107,806 transfers measured · 9 EVM chains and the XRP Ledger · data through 2026-09-27 · method · Source: Teiten Research
What the covered half looks like
Method — 7 deviations: §2.33 · §2.25 · §2.4 · §2.19 · §2.2 · §2.20 · scope
- The Brazilian settlement rail is inside these figures, not beside them, since 2026-09-25. On celo, BRLA settles a payment as a mint into a customer address and a burn from the same address seconds later, with no transfer between two addresses; §2.3 removes both legs from a transfer-based series, so until that date these figures reported zero for those payments. Each mint into an address that has never been either side of a peer transfer is now counted as settled value at the day's official rate and as one settlement in the transfer count, and each such address minted to or burned from in the period is an active wallet. The two populations are disjoint by construction — a mint is a row the peer side excludes, and a settlement address has no peer row — so nothing is counted twice. The class is fixed on full history and per contract, and the rail on its own, with its share, is published as the br-settlement dataset. (celo BRLA, 2026-01 to 2026-08: R$18,449.9M through the rail against R$13,892M of peer transfers on the same token — before this change the figure was at most half of what settled)
- SUPERSEDES the source these figures used to be built from. Until 2026-09-01 every US-dollar figure here was converted at a market series of unrecoverable provenance that ended 2026-02-05; it is now each central bank's own published rate — Brazil's PTAX closing, Mexico's FIX, Colombia's representative market rate, Argentina's reference rate — applied to the whole history rather than only to recent dates. A figure taken from this site before that date will not match the same figure now, and the exchange rate is why. The older series has not been withdrawn: it is still published as its own dataset, so the difference stays checkable. (restating settled value onto official rates: ARS +76.4%, BRL +0.7%, COP -0.5%, MXN +3.1%; $6.98bn to $7.61bn in total)
- Official FX, not market price: each central bank's own published rate, carrying that bank's own rate type. ARS carries the widest official-versus-parallel spread of the four currencies, so Argentine dollar figures understate what people transact at by that gap. The rates, their publishers and their licences are listed on the Official exchange rates dataset.
- RETIRED 2026-09-01, and it must not be reintroduced. MXN used to have no US-dollar value before 2024-06-25, because the old market series opened the peso at 10.01 to the dollar and held 10.01 to 10.18 for seven days against a peso that has traded 17 to 21 throughout. That was a defect of that series, not of the peso: Banco de Mexico's own FIX reads 18.45 on 2024-06-14 and 18.18 on 06-21, correct across the entire window the rule deleted. Those days are now published at the central bank's rate. (the 7 days this rule used to blank carry MX$6,378,724 of native volume, now valued at $350,996 — against the $588,925 the discarded rates would have invented)
- §2.2's disputed-decimals exclusion is not applied to amounts; §2.15 supersedes it. The `conflict` flag marks `decimals_metadata` blanket-claiming 18 decimals, not a wrong applied value: `tokens.decimals` is right in every conflict case except MXNB, which carries decimals = 0 and is corrected by a further 10^-6 here. Excluding the flagged rows deletes whole chains instead of adjusting them, and that is why Mexico used to read near-empty in value. MXNB and MXNe are priced, so Mexico has a value series. (at the week of 2026-01-26, bnb BRZ resolves to exactly R$100,000,000; excluded, BRZ reads R$74,982,627 against R$174,982,627 gross issued)
- Every MXNB figure, and every MXN figure containing one, is a lower bound rather than a measurement. MXNB trades on five chains and this database holds transfer rows for two of them, arbitrum and ethereum: `onchain.tokens` carries an avalanche_c entry with no rows behind it, and polygon and base have no registry entry at all. Arbitrum itself begins in November 2024, so MXNB's launch period is missing outright. It is kept rather than dropped, because deleting it would misstate MXN as badly in the other direction. (4,304 MXNB rows on arbitrum, none before 2024-11-07, and 635 on ethereum; avalanche_c registered with zero)
- Yield-bearing stablecoins (TESOURO, CETES) are reported separately and are not in these figures. (1,473 rows against a published count basis of 9,569,467 transfers — 0.0154%)