Our rates beside a public reference

Both sides cover the same window. We do not take the reference's headline number. We take their daily series, cut it to exactly the span our own figure covers, and average it over that span. Measured on one series in September, comparing our realized return against their current reading instead put the apparent gap at 0.45 percentage points; matching the windows put it at 0.02. The window mismatch was almost the whole difference.

We use their base rate, not their headline rate. Their headline figure can include reward-token incentives. Our index measures what the protocol itself paid. Comparing the two would set lending income against lending income plus rewards.

The pool is re-identified every time you load this page. A pool id is not self-describing, and four Aave v3 pools on Ethereum report the symbol USDC with materially different rates. Each pin records the project, market and symbol it must still be. If any of those changed, the row is refused rather than shown.

The two sides measure different things, so a small gap is expected and neither side is wrong. Ours is what actually accrued: a cumulative index read at two signed blocks, capturing every unit of income in between. Theirs is an average of daily snapshot readings, which cannot see what happened between snapshots. An average of samples is not the growth of the thing being sampled, so the two will not land on the same number even when both are correct.

That is why this page states the size of a gap and what could produce it, and never says who is right. Where our own series is missing days inside a window, the row says so, because their average then covers time our figure does not.

The reference side of this table is not signed and not attested. Your browser fetches it directly from a third party. It never enters a proof bundle, never reaches the figure we sign, and we cannot vouch for it. Agreement here is corroboration, not verification.

Our side is derived from two signed measurements. The rate itself is arithmetic we publish rather than something we sign, so the thing to check is the endpoints, and those you can verify without trusting this page or us. The feed serves them and the rates page shows the covered span for each series.