Methodology
How these numbers are collected, what they mean, and — the part most sites leave out — where they fall short.
Where the data comes from
Every figure comes from an official statistical source: central banks, treasury departments, and financial regulators. We do not collect rate quotes from lenders, because a rate a lender chooses to advertise to us is marketing, not observation. The full list, with attributions, is on the sources page.
How often it updates
The pipeline runs daily. That does not mean every number changes daily — sources publish on their own schedules, and we do not disguise the difference:
- Daily: government bond yields, policy rates, US locked-rate mortgage indices.
- Weekly: the US mortgage survey, Canadian posted rates.
- Monthly, in arrears: European bank interest rates (about two months behind), Canadian rates actually charged (about three).
- Annually: US loan-level mortgage data, roughly eighteen months behind.
Every page states the observation date of the figure it shows. If a number is from two months ago, the page says so rather than implying it is current.
Revisions and vintages
Economic statistics get revised. A figure published for March may be corrected in May. Every observation we store records both the date it describes and the date we learned it, so a revision adds a new record rather than overwriting the old one. Where a figure has been revised since we first saw it, the page shows its vintage.
A limitation, stated plainly: we began collecting in July 2026. Revisions made before then are not in our archive — for older observations we hold the currently-published figure, not the sequence of values that preceded it. The record of what was believed at the time builds forward from our start date, and we would rather say so than let the machinery imply a history we do not have.
What we do when something looks wrong
Every series carries plausible bounds. A 30-year mortgage rate of 0.4% is a parsing error, not news, and values outside those bounds are quarantined rather than published. Verification runs before publishing, not after: if a check fails, the previous day's figures stay up. A stale number that is correct is better than a fresh one that is wrong.
Missing data is shown as missing. We never interpolate a value and present it as an observation, and where two series of different frequencies are compared, each point pairs figures that were both genuinely published at that time.
What this site cannot tell you
These are market-level statistics, not quotes. An index reflects what a population of borrowers paid; the rate you are personally offered depends on your lender, your finances, and your timing. Nothing here is a prediction, and nothing here is advice. We report what rates are — never what you should do about them.
Cross-country comparisons carry a further caveat. A thirty-year American fixed mortgage, a five-year Canadian fixed, and a German loan with a ten-year fixation are genuinely different products. The comparison is meaningful, but it is not like-for-like, and we would rather say so than let the table imply otherwise.
Corrections
When we get something wrong we fix it and record it publicly on the changelog. If you find an error, that is a service to everyone reading — please report it.
Current coverage
272 series across 25 countries and areas, 249,063 observations, data current to 2026-09-14.