PropRaven Marketplace v1
Canonical real estate intelligence for 191.3M U.S. parcels — resolved owners, transactions, loans, distress signals, and per-peril risk — delivered as Snowflake shares.
Quick start
- Listing status (2026-10-15). The Snowflake Marketplace listing is in review, so it cannot be subscribed to yet.
- Access today. Access is a direct share by request: email sales@propraven.com with your Snowflake account identifier and the listings and states you need. It is not self-serve yet.
- Requirements. The provider account runs on AWS (region on request). To accept the share you need a role with
IMPORT SHAREandCREATE DATABASE(standard Snowflake share acceptance), so you can create a database from the share. - Cost. Snowflake delivery is sold as a Data License; see /pricing.
- Email sales@propraven.com for a direct share. A good starting point is Preview (free) or Parcel Intelligence for the wide table.
- Accept the share in your account by creating a database from it.
- Run sample queries against the share in Snowsight.
Catalog
Nine listings, three access modes: a wide-table for engineers, a knowledge-graph for analysts, and pre-filtered vertical cohorts for specific buyers. Same underlying truth.
AVM, FEMA NRI, demographics, HMDA, crime, climate, permits, deeds, owner-entity resolution. One row per parcel. Best for engineers integrating a single feature into an existing workflow.
12 canonical entities, 166M+ ownership edges, confidence-scored signals. Designed for graph traversal, entity resolution, and entity-model app development. Same truth as Parcel Intelligence, different access shape.
12.6K recorded loans (feed being re-expanded) + canonical lenders + per-lender exposure rollup with distress overlay. Pre-joined to property + owner context.
Per-parcel multi-peril risk vector — FEMA NRI rated (~92% coverage), NFIP claim history, NOAA storms, USGS seismic, CDC SVI, wildfire.
Every U.S. bank branch with deposits, ownership chain, geocoded location. Pre-computed branch stability scores, MSA market share, antitrust HHI.
183M residential parcels (SFR / MF_SMALL / CONDO / MOBILE) with absentee classification, institutional flag, recent transfer.
11M commercial / industrial / multifamily / mixed-use parcels with owner graph and intelligence signals.
191.3M parcels with addresses, property attributes, tax and assessment fields where the county publishes them, 157M+ permit records, and the full Census geographic hierarchy. No owner data.
Free 10K-parcel sample of every entity. Same schema as the paid catalog. No commitment required to evaluate.
Reference
- Data dictionary — every column on every entity.
- Methodology & versioning — stable IDs, confidence model, refresh cadence.
- Sample queries — ready-to-run SQL across the catalog.
- Databricks Delta Sharing — the
propraven_silvershare, self-service provisioning, and Databricks / Python / PySpark snippets.
Refresh cadence
v1 listings are refreshed per parcel build. PROPRAVEN_PARCEL_INTELLIGENCE (the 620-column parcels table) was refreshed on 2026-10-15 from the parcels gateway's current per-state builds (most states 2026-10-12; ID 2026-10-11; CT 2026-10-04-rw1; WI 2026-10-18), loaded by serverless Snowpipe. The Ontology share's silver tables were refreshed the same day from silver built on that per-state map; its derived graph and intelligence tables (owner portfolio, ownership edges, distress, value-add, institutional ownership, lender exposure) are from the 2026-08-28 build. The build currently served by the API is shown on /data/live. The catalog has 620 catalogued columns (44 rebuilt every epoch and 576 carried from an April 2026 reference export; 56 of the 620 are withheld from sale while they are rebuilt). Source data inherits its publishing cadence; per-row _LOADED_AT and _SOURCE_AS_OF columns expose actual freshness. See methodology for the full breakdown.
The API's GET /api/v1/freshness reports two different clocks. content_as_of is the content date of the served data: how current its records are, measured as content_date_basis names and content_date_note explains. swapped_at and snapshot_built_at are the last national ratify, when all states were last published together on one release. A swap re-serves data without making it newer, so swapped_at can be weeks later than content_as_of. Postgres serves parcels per state, and states move to new releases on different days, so the payload's parcels_pg.per_state gives each state's release and activation time and parcels_pg.summary gives the national spread (oldest, newest, states per release).
Support
Questions, custom share requests, or coverage requirements that don't fit a published listing — email support@propraven.com.