How the yield in a yield-bearing stablecoin is measured

TESOURO and CETES are currency-pegged and also yield-accruing: each is a tokenised claim on government debt, Brazilian Tesouro Direto and Mexican Cetes. They are a different instrument from the payment stablecoins in the headline series — someone moving BRZ and someone moving TESOURO are doing different things — so they are reported here rather than folded into a figure that would merge a payments signal with a yield signal.

Until September 2026 that separation rested on a suspicion. The tokens are named after debt instruments, so they were assumed to accrue, and this page said so and withheld any value on that basis. Assuming is not measuring, and the assumption was recorded as unverified for as long as it was one.

It is measured now, and not from a document — from this database's own record of swaps. Each token is priced on its own order book against a counterparty that is pegged one-to-one, and that price is divided by the token's currency at its central bank's official rate. A token trading exactly at its peg gives zero. TESOURO went from 17.6 to 26.0 per cent above par across the twenty-two published weeks ending 2026-08-24, rising in thirteen of twenty-one steps, which annualises at 14.8% — against a Brazilian policy rate near 15%. CETES went from 13.8 to 16.3 per cent over seven published months, an accrual of 3.0% a year, which is well below what a Cetes bill pays; this is what the pool traded at, and no attempt is made here to reconcile the two. The Stellar network's own order book puts CETES 18 per cent above par independently, on infrastructure that shares nothing with the Ethereum-family venues.

Two guards stand between a price and an arithmetic accident, and both leave gaps rather than numbers. Each side of a trade must be worth at least ten dollars — a rule that removes 477 of CETES's 727 raw observations, and with them rows whose near-zero side divided out to 106,899,124 pesos per unit. And a period publishes no price at all unless five trades clear that bar, because below it one participant's round trip is the whole reading. Together they moved the worst published CETES week from 419 per cent above par to a gap.

So the premium chart is the yield, and it is deliberately full of holes: it exists only where the token actually traded enough in that period, and is never carried across a quiet one, because holding the last ratio forward would draw an accrual nobody paid. CETES is thin enough that only four weeks in its whole history clear the bar, which is why the same series is published by month as well.

Crown's BRLV is shown here as the stable leg of a yield-bearing product, and the same measurement is what keeps that honest. Every one of its twenty-two published weeks falls between 1.5 per cent below par and 1.6 per cent above it, and the trend has no sign — the weekly series annualises to +2.5% and the monthly to −0.9%. Two grains disagreeing about the direction is what no drift looks like. It is on this page because the yield in Crown's design sits in a second token, BRLY, which this database does not yet hold. Its flat line beside TESOURO's climb is the difference between the two, drawn rather than asserted.

Every dollar figure on this page is a floor, and that word is doing real work. Units are converted at the currency's official rate — the same arithmetic every other token on this site is valued by — and for a token that is a one-to-one claim on its currency, that is the value. For one that accrues it is the low end: at the last week each token was priced, the floor understates TESOURO by $149,491 of $725,471 and CETES by $687,684 of $4,702,688, 25.9 and 17.1 per cent. Crown's BRLV, which is pegged, is understated by 0.9 per cent, which is measurement error rather than a yield. Until 2026-09-03 no dollar figure was published here at all, and the change is what is valued rather than a change of mind about the caveat: the floor exists so that three tokens in two currencies can be compared on one axis, which is a question this page could not previously answer.

The premium is not applied to the floor for you, and the market-valued stock is not published at any aggregation. Both follow from the same fact: balances are known every calendar day and prices only in periods that traded enough to publish one — 22 of TESOURO's 66 published weeks, 22 of BRLV's, four of CETES's. Applying the premium would mean carrying a ratio across the two thirds of weeks that have none, which draws an accrual nobody paid; multiplying units by price would give a stock with holes in it, vanishing on a quiet week and returning unchanged as though something had been redeemed. The two series are published side by side and multiplied by a reader who wants a point estimate, in the periods where both exist.

Balances include the Stellar network; transfer counts and settled value do not. Every Stellar row held here touches an issuing account, because the sync that has run follows the issuers themselves — which sees every change in supply and no transfer between two holders. A Stellar transfer count would therefore be a hard zero on an asset that is demonstrably live, and unmeasured is not zero.