Where the data came from
In effect UNSET — effective date
Every building on the board is a real building and every rate in the market is a rate that was really charged. That is only worth claiming if the sources are named, so here they are, with the terms each one is used under.
Building footprints, street geometry and addresses are © OpenStreetMap contributors, available under the Open Database Licence. The map you play on is drawn from it.
The map files this product ships — austin-tx.json, chicago-il.json and phoenix-az.json, served from /data/map/ — are extracts of the OpenStreetMap database taken through the Overpass API. They are a Derivative Database, not a rendering of one, and they are offered to you under the ODbL 1.0 on the same terms we received them. Download them directly: Austin, Chicago, Phoenix.
The streets, water and blocks beneath the footprints are vector tiles we cut ourselves from the Protomaps daily planet build and serve from our own storage. The cartography is Protomaps, under a BSD licence; the data underneath it is © OpenStreetMap contributors, under ODbL, the same as everything else on the board.
The market history the simulation runs on comes from FRED, published by the Federal Reserve Bank of St. Louis: the effective federal funds rate, the Consumer Price Index, the unemployment rate, the rental vacancy rate and median household income. Each is a work of the United States government and carries no copyright.
Mortgage rates — 30-year fixed, 15-year fixed and the 5/1 ARM — are the Primary Mortgage Market Survey, © Freddie Mac, used with attribution. Rates between survey dates are interpolated for the simulation; the survey itself is unaltered.
House prices are the FHFA All-Transactions House Price Index — the national series, and a separate one for each city, because Phoenix and Austin did not have the same decade and a national average says neither. The neighbourhood view also uses FHFA’s annual index for each census tract, to carry an area’s recorded median home value back to the season’s year. Published by a United States government agency and free of copyright.
With real prices on, a building’s type, year built and value come from the appraisal district’s parcels, published by Travis County for reference purposes only. The parcels are kept as a separate layer from OpenStreetMap. The game matches them while it runs: each building belongs to the parcel its footprint’s centre falls in, or the nearest within 15 metres; on a parcel with several buildings the largest is the property and much smaller ones are outbuildings, not for sale; buildings of similar size share the parcel’s value by footprint area. A building’s homes are its census tract’s count of homes (American Community Survey 2019–23) shared across its buildings by type and footprint. Nothing joining the two is stored or published.
The map’s optional relief shading is drawn from Terrain Tiles, open elevation data compiled by Mapzen and hosted by Amazon Web Services as part of its Open Data programme. For the cities here the underlying elevation is the US Geological Survey’s 3D Elevation Program, a public-domain work. Nothing is fetched unless relief is switched on.
Neighbourhood character is read from the American Community Survey by census tract, located through the Census geocoder and the FCC area API. Both are federal services and neither is copyrighted.
What is recorded, and what is not
The geometry, the addresses and the market history are recorded fact. Everything attached to an individual building — its price, its condition, its rent, its tenants — is modelled by this simulation from the metro’s profile. None of it is a listing, an appraisal, an automated valuation or a statement about the property or the people who own it. Do not use it as one. The terms say the same thing in the place it is binding.
Corrections to anything on this page are welcome, and corrections to the map itself belong upstream at openstreetmap.org, where everyone gets them.