How we compute the real rate

Unposted answers one question: for a mortgage like yours, what are people actually signing for? Here is how we get there.

Sources

We collect rates that borrowers publicly report getting, in Canadian mortgage communities and forums where sharing your deal is the norm. Each observation records the rate, term, type (fixed or variable), the insured tier, the deal (purchase, renewal, refinance, switch), and region where the borrower gave them. We separately track lenders' posted and advertised special rates, checked directly, and the Bank of Canada's published market series (posted mortgage rates, bond yields, the policy rate) as an independent anchor.

Your cohort

When you check your rate, we find offers that match your profile, the same product, insured tier, deal type, and region, and report the median of that cohort along with its middle range. The median is the typical signed rate; the low end (10th percentile) is what the sharpest shoppers reached. If your exact cohort is too thin to read precisely, we widen the match one step at a time and tell you we did, rather than pretend to a precision the data cannot support.

Lenders vs channels

Our lender tables rank originating lenders (the banks, monolines, credit unions, and private lenders that actually fund mortgages). Brokers and rate marketplaces are a channel that arranges a mortgage with one of those lenders, not lenders themselves, so we keep them out of the “who lends” ranking and look at the broker channel separately.

The rules we hold ourselves to

• Every number is real. Each figure traces to actual reported offers or published market data. We never invent or round toward a more flattering story.
• Sample size and date, always. Every rate is shown with how many offers it rests on and when the newest one landed. Thin samples say so.
• Achievable, not average. People who share their rates skew toward engaged shoppers, so our benchmark reflects what a motivated borrower can attain, and we label it that way rather than calling it a market average.
• We widen honestly. When there is too little data for your exact cohort, we relax the match step by step (city to province to national, then to a model) and tell you which step we used, instead of showing a precise-looking number built on almost nothing.

What this is not

Reported rates can be mis-entered or out of date, and a self-reported sample is not a random one, so treat every figure as a well-sourced estimate rather than a guarantee. Unposted is an information tool, not a mortgage offer or financial advice. Terms are defined in the glossary.