Methodology, coverage and limitations

What the numbers are

Every figure is computed from observed asking rents on public rental listing pages. RentScope never claims signed-lease prices, transaction prices or investment returns. Listings are normalised, quality-checked, deduplicated and canonicalised; only eligible listings observed within the last 7 days enter statistics. A metric is published only if its sample, confidence, freshness, quality, coverage, methodology and source-control gates pass. Unsupported numbers are replaced with an explicit suppression notice — never with a weaker estimate.

Freshness and availability

RentScope reports when it observed each listing, not whether the property is available right now. “Last observed by RentScope” is the most recent successful check; “source published” and “source expiry” are the listing's own dates from the source. A passed source expiry does not by itself mean the listing is gone — owners renew and re-post listings between observations — so RentScope never treats expiry, a missed crawl or a failed request as proof that a listing was rented or removed. Availability can change at any time and RentScope does not guarantee real-time availability.

Methodology versions

Normalisation

normalizer-1.5.0

Geography resolver

geo-resolver-1.1.0

Building resolver

building-resolver-1.1.0

Deduplication

dedupe-1.1.0

Comparables

comparables-1.0.0

Fair rent

fairrent-1.2.0

Deal Score

dealscore-1.1.0

Market analytics

analytics-1.0.0

Publication policy

RentScope publishes aggregated analytics and normalized listing facts. It never publishes seller names, phone numbers, email, Zalo or any contact detail (none is collected), and never exposes raw source payloads or internal evidence. Sources can be switched off at runtime (acquisition, analytics, publication); switching a source off removes its listings from every statistic immediately.

Source and publication policy: RentScope aggregates and analyses rental-market information from supported sources under its current publication policies. Publication in RentScope does not imply endorsement, partnership, affiliation, or permission from the original source. Source provenance is retained where applicable, and private contact information is not published.

Known limitations (stated, not hidden)

• Single source. Current statistics come from one public rental platform (Phongtro123). Coverage is partial and biased towards rooms and mid-market apartments.
• Building resolution is conservative (~46%). Unknown is preferred over wrong, so most listings have no building page. Building medians therefore describe only the resolved subset.
• Confidence labels have two different meanings. “Sample confidence” on aggregate pages (market, areas, buildings) describes how much data supports a statistic. Published Fair Rent estimate confidence is capped at MEDIUM: on real data HIGH has not yet proven more accurate than MEDIUM (fairrent-1.2.0), so no HIGH estimate label is published. Deal Scores are capped at 60 for LOW confidence.
• Fair-rent coverage is ~45%. Many listings lack enough comparable data; those show “—”, never a fabricated estimate.
• No history features yet. RentScope has observed this market for fewer than 30 days; price-drop and days-on-market features require 30 days and trends require 90 days of real observations. Days-on-market, price-drop alerts and trends are withheld.
• Deal Score is relative, not advice. It measures asking rent against comparable listings, nothing else — not quality, not negotiation, not returns.
• Availability is not guaranteed. RentScope observes listings and reports when it last saw them; it cannot yet confirm from the source when a listing has been rented or removed, so availability can change between observations.

Privacy

Listing contact details are removed before storage. RentScope stores listing facts, public source URLs and observation timestamps. Price-check inputs are hypothetical queries and are never stored as listings or mixed into market statistics.