The BrightCat Home Price Index measures Canadian residential appreciation using repeat-sale methodology — the same property tracked across multiple confirmed transactions. No composition bias. No smoothing artifacts. No estimates.
Average-price indices smooth real movement until you can't see it. Repeat-sale shows you what actually happened.
A direct measurement of Canadian residential property appreciation derived from confirmed sale-pair data, not from listing snapshots or assessor valuations.
The methodology was popularised by Case & Shiller for US property indices in the 1980s and has since become the academic standard for measuring residential appreciation without compositional distortion.
A repeat-sale index requires three inputs: a persistent property identifier that tracks the same property across multiple transactions, confirmed transaction prices for each sale, and a confirmed date for each sale. With those three inputs, the appreciation between any two sales of the same property can be calculated directly. There's nothing modelled, nothing estimated, and nothing inferred from comparable properties.
The mathematics is straightforward. For each repeat-sale pair, the appreciation factor is the second sale price divided by the first sale price. Aggregating across thousands of pairs in the same geography and period gives a robust measure of how prices have actually moved on a like-for-like basis — without the distortion introduced when the mix of properties selling in a given quarter happens to be heavily weighted toward one segment.
What this method requires — and what most Canadian data sources cannot provide — is property-level identifier persistence across transactions. MLS numbers do not persist; they're assigned per listing, change on every relisting, and aren't unique across regions. Provincial registry records persist by parcel, but registry data lags real market activity by weeks to months and lacks listing context. BrightCat's persistent property identifier system was built specifically to solve this problem — joining listings to sales to relists to subsequent transactions on the same property across years.
Why a property-level index reads differently from the headlines.
Each row in the BrightCat HPI dataset is a single repeat-sale pair — one property, two confirmed transactions.
The renovation flag identifies pairs where the lifecycle between sales suggests material property changes — extended delistings followed by relistings at notably higher prices — useful for filtering true price appreciation from value-add appreciation. Full schema with type definitions, enums, and example values available to licensed clients.
Four recurring buyer profiles for repeat-sale data.
Repeat-sale only works if the matching is right. A wrong match can fabricate appreciation that didn't happen.
BrightCat's pair-matching pipeline rejects more candidate pairs than it accepts. The persistent property identifier joins on standardised address, parsed unit number, postal code, and property characteristics — not on MLS number, which fragments across relistings. Pairs where the second sale shows a substantial physical change (square footage, bedroom count, building type) are flagged as potential renovation cases and either excluded from the headline series or marked for downstream filtering. Pairs with implausibly short time gaps and large price jumps are reviewed before inclusion.
The result is a conservative pair set rather than the largest possible pair set. We prioritise pair integrity over pair count because a fabricated pair introduces more error than a missing one. Full methodology documentation is available for buyers running model risk reviews, regulatory data lineage assessments, or AI governance audits.
A repeat-sale home price index measures property appreciation by tracking the same individual property across two or more confirmed sale transactions. The methodology was popularised by Case & Shiller in the 1980s and has since become the academic standard.
As of April 2026, BrightCat holds 194,167 confirmed repeat-sale pairs derived from 12 years of continuous Canadian transaction capture since 2014.
Average-price and median-price indices are distorted by composition — the mix of properties that happen to sell in any given period. Repeat-sale removes this by comparing each property only to itself.
Yes. Repeat-sale pairs are the gold standard training data for automated valuation models because they provide ground-truth appreciation rather than modelled estimates.
Both use repeat-sale methodology. Teranet–National Bank HPI is constructed from provincial land registry data; BrightCat HPI is constructed from BrightCat's continuous transaction pipeline which also preserves the full listing lifecycle leading up to each sale. The two are complementary reference series.
Via Snowflake Marketplace (Secure Data Share), MCP connector for AI agents, or structured flat files. Sample data is free; production access is governed by an annual Master Data License Agreement.
Sample repeat-sale data covers Greater Toronto, Greater Vancouver, Greater Montreal, Calgary, and Ottawa. Verify the pairs in your models before scaling.