Follow the evidence back to its source.

Connect the national picture with local evidence

Research starts with Aurynge's crawled archive, place-based briefings and local narratives. Source checks then follow the underlying records and current releases from statistical agencies, the Bank of Canada, housing organisations, real estate boards and public authorities. When an industry report supplies local figures, the article identifies it as such.

The analysis looks for a concrete connection: how a financing change could affect a development, how the mix of completed homes compares with buyer needs, or how new housing in one community could change choices in another. A national trend is context, not proof of the same trend in every community.

Articles are prepared with AI-assisted research and writing. Interviews, site visits and named expert reviews must be real and attributable. The byline identifies Aurynge as the publisher; it does not imply that a public authority or cited organisation endorses the analysis.

Keep four dates separate

A document's publication date, the period its figures describe, its retrieval date and the article's publication date can all differ. Each article records its publication and source-check dates. Source notes flag historical material and future project estimates. Finding an old planning page today does not make it a new announcement.

Match the geography and the units

Articles distinguish property-level and local evidence from municipal, regional, provincial and national context. A wider-area forecast cannot establish what a particular home will sell for. Homes, bedrooms, permits, starts, completions, rental units and ownership units are different quantities. Those differences remain explicit wherever they affect the interpretation.

Separate observations, calculations and forecasts

Reported figures carry a source and reference period. Calculations identify their inputs and assumptions. Interpretation explains how those observations fit together; it does not establish statistical proof that one event caused another.

Forecast sections state the expected direction, the time horizon, the underlying assumptions and the evidence that would change the outlook. These editorial outlooks are reasoned scenarios, not outputs from a fitted price-prediction model. Numerical probabilities and price targets require a supporting method.

Housing targets, planning approvals, construction starts and occupied homes describe different stages. Market benchmarks also differ from sale averages and individual valuations. These distinctions remain visible where they change the conclusion. Interactive calculator results depend on the inputs and evidence shown with the result.

Check the archive before trusting a summary

A fresh crawl can contain an old forecast, and several summaries can repeat the same underlying source. Source checks cover publication periods, revisions, geography and original figures before an article uses them. Conflicting or incorrectly labelled statistics are excluded when they cannot be reconciled. A large archive improves discovery; its size alone does not validate a claim.

Publication and corrections

Published articles are dated editions. New source records do not automatically rewrite them. An update must carry valid sources and publication dates before it replaces an edition. A changed date should reflect a substantive update to the article.

To report a correction, email brad@aurynge.com with the article URL, the disputed statement and a supporting source. Material corrections will be explained in the article when an updated edition is published.

Measure what changed in the accumulated records

Archive comparisons preserve the earlier and later observations, match the same record identifiers, and include the inputs and calculations alongside the article. Repeated crawls of unchanged content do not count as new development. A missing record has an unknown outcome unless another source explains it.

A local planning record can matter even when it is unsuitable for a national market summary. If a relevance filter excluded it, the evidence record retains that reason and documents a separate source check before the article uses it for local analysis. Neither the classifier nor its override proves a market trend.

Make predictions that can be checked

Original, data-based hypotheses are part of the research. A proposed insight must explain what the evidence adds beyond ordinary market knowledge, who it affects locally and what would contradict it. General principles, such as the need to finance construction, belong in the background. Missing evidence of a local effect remains a research question. A useful forecast explains the causal chain, competing forces, assumptions and deadline. Selected forecasts include a fixed cohort and scoring rule saved before the observation period. Later checks report whether the forecast was met, missed or could not be resolved from the available records.

Recording a forecast does not establish that an early indicator predicts later housing statistics. Demonstrating that relationship requires comparable release dates, a fixed cohort, repeated outcome checks and recorded failures. One correct prediction does not establish a reliable forecasting model.