Canadian property data for mortgage lending, AVMs, and REIT analysis

Mortgage risk models, automated valuation models, home price indices, and REIT portfolio analytics all depend on the same underlying asset: Canadian property data. But the data requirements differ by use case, and the Canadian market has structural gaps that make sourcing harder than in the U.S. This page maps the data inputs to the use cases — what you need, where it comes from, and what's actually available.

Mortgage risk: what property data lenders actually need

Canadian mortgage lenders use property data at two points: origination (is this property worth what the borrower says it is?) and portfolio monitoring (are the properties backing our book still performing?).

At origination, the critical inputs are comparable sold transactions (what have similar properties in this area sold for recently?), active listing context (what is currently on the market in this neighbourhood and at what price?), and property characteristics (type, size, age, condition). Together, these allow a lender to validate the appraised or declared value of the collateral.

For portfolio monitoring, the inputs shift to signals: are properties in the areas where we hold mortgage exposure showing distress? Distress signals include rising days-on-market, increasing price reductions, growing inventory relative to sales, and properties being relisted after failed sale attempts. These signals appear in listing data weeks or months before they show up in transaction records or default statistics.

The gap in the Canadian market is that most lenders rely on appraisal data (backward-looking, property-by-property) and assessment rolls (annual, lagging). Weekly listing and sold data provides a real-time layer that traditional sources miss. See property data for banking for BrightCat's banking-specific applications.

Mortgage default prediction: the data inputs

Mortgage default prediction models combine borrower-level data (credit score, income, debt ratios) with property-level data (collateral value, market conditions). The property side is often the weaker input because it depends on static appraisal values rather than current market signals.

Property data improves default prediction in three ways:

• Negative equity detection. If the current market value of a property has fallen below the outstanding mortgage balance, the borrower is in negative equity — the single strongest predictor of strategic default. Detecting this requires current comparable data, not the appraisal from origination.
• Market stress signals. Rising inventory, falling prices, and increasing days-on-market at the postal code level indicate market stress. Properties in stressed markets carry higher default probability even when borrower fundamentals look acceptable.
• Property lifecycle anomalies. A property that is listed for sale while the mortgage is still active may indicate financial distress. A property that is listed, pulled, and relisted multiple times may indicate difficulty selling — which matters if the lender needs to recover collateral.

Weekly property data feeds make these signals available for portfolio-level monitoring across thousands or millions of mortgages simultaneously.

AVM data sources in the Canadian market

Automated valuation models estimate a property's market value without a physical appraisal. The quality of the estimate depends almost entirely on the quality of the data feeding the model. In Canada, AVM data comes from four sources:

Sold transaction data is the foundation. Every AVM methodology — comparable sales, hedonic regression, repeat-sale — requires historical sale prices. The more recent, geographically precise, and comprehensive the sold data, the better the model performs. In Canada, sold data availability varies by province due to disclosure restrictions, which is why independent providers that aggregate across sources are critical for national AVM coverage.

Active listing data provides real-time market context. Listings show where the market is heading, not where it has been. List-to-sale ratios, days on market, and inventory levels are calibration signals that improve AVM accuracy in fast-moving markets. See why listings are the earliest signal.

Property characteristics from assessment rolls (bedroom count, lot size, building age, construction type) provide the feature set for hedonic models. These are relatively static and widely available from provincial assessment authorities.

Repeat-sale price pairs are used specifically for constructing home price indices and for AVM methodologies that track the same property over time. A repeat-sale pair requires two or more transactions for the same property, linked by a persistent property identifier. See AVM data in Canada for a deeper dive.

Building a repeat-sale home price index with Canadian data

A repeat-sale home price index measures price changes by comparing the same property's sale price across two or more transactions. This approach controls for property characteristics (because you are comparing the same house to itself) and isolates pure price appreciation from composition effects.

The data requirements for a repeat-sale index are specific:

• Sold transaction records with sale prices and dates.
• A persistent property identifier that links the same physical property across transactions, even if the address formatting changes, the property is subdivided, or the owner changes. Without a stable identifier, the same property cannot be matched across sales.
• Sufficient pair volume — the index requires enough repeat-sale pairs in each geography and time period to produce statistically meaningful estimates. Thin markets with few repeat transactions produce noisy indices.
• Quality filtering — pairs must be screened for non-arm's-length transactions, flips (where renovation materially changes the property), and data errors. Unfiltered pairs introduce bias.

BrightCat's Core product includes the BrightCat Home Price Index, constructed from repeat-sale pairs across the national sold dataset. The underlying methodology is documented on the price index page. For more on methodology, see home price index methodology.

REIT portfolio analytics: what data feeds the analysis

Real estate investment trusts use property data for four core functions:

Portfolio valuation. REITs must mark their holdings to market value. This requires comparable data — recent sold transactions and active listings for similar properties in the same markets. The more granular and current the data, the tighter the valuation.

Acquisition screening. REITs evaluating potential acquisitions need property-level data to identify candidates: properties of a specific type, in a target geography, within a price range, with specific characteristics. Screening at scale requires structured, queryable data — not PDF reports.

Market monitoring. REITs track the markets where they hold exposure. Rising inventory, falling rents, or increasing days-on-market in a market where the REIT has significant holdings is an early warning. Weekly data makes this monitoring actionable rather than retrospective.

Disposition timing. When a REIT decides to sell a property, the listing data for comparable properties determines the optimal pricing and timing strategy. Properties listed during periods of low inventory command stronger prices than those listed into a rising supply wave.

For a detailed treatment, see property data for REIT portfolio analysis.

What data is available for these use cases in Canada

The Canadian data landscape for mortgage, AVM, and REIT use cases is thinner than the U.S. equivalent. Here is what exists:

• Provincial land registries (Teranet in Ontario, LTSA in BC) provide title and transaction records. Strong within their jurisdictions but no national coverage and no listing data.
• CREA / MLS provides listing and transaction data through member boards. Comprehensive for MLS-listed properties but access-restricted and not available as a raw data feed for non-member enterprise use cases.
• Assessment authorities (MPAC, BC Assessment) provide property characteristics and assessed values. Annual updates, useful for AVM feature inputs but not for current market signals.
• Independent providers — BrightCat Data provides national coverage across listings, sold, rentals, commercial, and core datasets with weekly updates. The combination of sold transaction data, active listings, and repeat-sale pairs in a single platform is what makes AVM calibration, mortgage monitoring, and REIT analytics possible at a national scale.

BrightCat's data for mortgage, AVM, and REIT use cases

The relevant BrightCat products for these use cases are:

• Sold — completed residential transactions with sale prices and dates. The foundation for AVM calibration and comparable analysis.
• Core — persistent property identifiers, repeat-sale price pairs, the BrightCat Home Price Index, geocoding, and enrichment. The linkage layer that connects transactions across time.
• Listings — active residential listings with lifecycle tracking. The real-time market signal layer for portfolio monitoring and disposition timing.
• Rentals — rental listings for income-approach valuation and investment property detection.
• Commercial — commercial property data for REITs with commercial holdings.

All products are delivered through Snowflake Marketplace and the BrightCat MCP connector. Explore the methodology for details on data construction and verification.

Frequently asked questions

Related reading