shinnslist · how we grade ourselves

How we grade ourselves

Name: shinnslist parcel coverage: 9 Colorado counties
Creator: shinnslist

Bottom line: three public rules. A persona opens only after its top results are checked by hand against the county’s own records. A signal counts as proven only after a test against its live public source passes. The ranking of who will sell is frozen every day and graded against what actually lists, and the grade is published even when it is zero.

Free beta 9 of 10 personas open · 263 of 313 signals proven · 2 days graded · as of October 19, 2026

The three rules

What has to be true before we say it

1 · Personas	A persona (a trade door on the hunt) is one query over recorded county facts, never a modeled estimate. It opens only after its top results were checked by hand against the county’s own records, and the check is dated. Today: Wholesaler (Top 12 checked by hand against Boulder County’s own records on Oct 16, 2026: sale, value and mailing all match, 12 of 12.) STR investor (Top 10 checked by hand on Oct 16, 2026: the owner’s mail goes elsewhere per Larimer County, 10 of 10; no sale in the county’s seven-year record, 8 of 8; the Big Thompson River within 150 m, 6 of the 6 the USGS answered for. Ten Boulder County rows: mailing and value match, 10 of 10.) Realtor (Checked by hand on Oct 16, 2026 against the counties’ live records: the 22 estates (an estate is the owner of record, the purchase, the latest recorded document, 22 of 22 each), the 10 code cases (the case outline holds the home, its type, status and date, and the purchase, 10 of 10 each), 12 of the 68 stalled permits (the permit holds the home with its kind and month, and the purchase, 12 of 12 each), and 12 of the 383 estate records (on the county’s live sales layer the document is on the account that day without a price and nothing priced follows it, 12 of 12; 4 of them on the Clerk and Recorder’s index: 3 named as the card says, 1 death record indexed as a supplemental affidavit). The top 12 of the 67 homes on Larimer County’s tax lien sale list, Oct 16, 2026: on the list for the amount the card says, the county’s address, and no sale of its own in the county’s seven-year record, 12 of 12 each. The owner of record as an estate or its heirs in every county, Oct 18, 2026: 12 drawn in Jefferson, Weld, Broomfield and Adams read so on the state’s live parcel record (each county’s owner of record as sent to the state, March and April 2026), 12 of 12.) Flipper (Top 12 checked by hand on Oct 16, 2026 against Boulder County’s live records: year built, size, county value and purchase, 12 of 12; all 65 sales next door found with the same date and price, and the median recomputed from them, 12 of 12. Adams County, Oct 18, 2026, 12 drawn: land and total value on the county’s live value layer 12 of 12, the purchase on its live sales layer 12 of 12, year built and size on its live improvements layer 12 of 12, and the first five sales next door of each, 60 of 60, found as deeds on those dates. 3,266 Adams homes on Oct 18, 2026.) Developer (Top 12 checked by hand on Oct 16, 2026 against Boulder County’s live records: land and house values, year built and size, lot acreage, the purchase, and the zone at the lot’s center, 12 of 12 each. Other counties, Oct 18, 2026: 12 drawn per county (all of them where fewer) against the county’s live record, land value, total value, year built and size: Weld 9 of 9, Adams 10 of 12 (2 a dollar apart), Jefferson 11 of 12 (one lot’s larger house is its second structure), Broomfield 1 of 1. Larimer’s and Eagle’s live records sit behind a disclaimer or a sign-in page, so they were not checked live.) Cash investor (Top 12 checked by hand on Oct 16, 2026: the Census Bureau’s own geocoder puts each home’s point in the census tract on its card, and the rent on the card is that tract’s in the survey’s published table, 12 of 12. 13,367 homes on Oct 16, 2026.) Fix and rent (Top 12 checked by hand on Oct 16, 2026: the Census Bureau’s own geocoder puts each home’s point in the census tract on its card, and the rent on the card is that tract’s in the survey’s published table, 12 of 12 (the dated-home facts as checked for the Flipper door). 1,025 homes on Oct 16, 2026.) Land (Top 12 checked by hand on Oct 16, 2026: the 8 Weld lots against the state’s public parcel composite (acreage, value and “Vacant Land”), 8 of 8; the 4 Boulder lots against Boulder County’s live assessor record (no building, the same land value and acreage), 4 of 4.) Family compound (Checked Oct 17, 2026 against Boulder County’s buildings file: 86% of the aerial’s second buildings on home lots (9,010 of 10,458) are on the county’s record as a garage or another building; the other 14% are buildings the record doesn’t list. The aerial finds 67% of the county’s detached garages of 400+ sq ft; the rest come from the record itself.) The other 1 stay closed until their data is real.
2 · Signals	A signal is proven when an acceptance test passes: labeled test cases, some that should fire and some that should not, graded against the signal’s live public source, with precision and recall of at least 0.90, with at least 80% of the test cases answered. The tests rerun every Monday. Today 263 of 313 are proven; the rest are built but untested, half-built or planned.
3 · The ranking	Every day the engine freezes its ranking of who is most likely to sell, so it cannot be changed later, and grades it against the homes that list afterwards. The ranking of who is most likely to sell has been graded on 2 days so far; the record is on How we grade.

What we never do

Buy data: every source is free and public.
Invent a count, a review or a testimonial. Every number on these pages is read from the records at each deploy.
Publish owner names or mailing addresses for search engines or AI crawlers: robots.txt closes every page and file that prints them.

The full record, one row per signal with its verdict and evidence: the signal truth table. The daily grade: the scorecard. What is open and what is not: the beta page.

When we are wrong

Reports, fixes and removals, counted in public

Every card says when its record was last checked and has a This is wrong button. Every report is checked against the county’s own record and either fixed or answered. Owners can ask for a home’s photo and owner details to come off at the removal page, done within 7 days. The Help button on every page reaches a person within a day.

Reading the record…

Proven so far

263 signals, with their tests

Best Value Number: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales, buildings, values and appraisal neighborhoods (the adapter from the public map services; the label from the nightly Sales.csv qualified sales, Buildings.csv, Values.csv and the parcel layer's NbhdCode read whole) and the City of Boulder's declared costs for single-family alterations.
Rent Potential Calculator: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against HUD Small Area Fair Market Rents and the Boulder MSA Fair Market Rents (HUD map services), Census ZIP areas (TIGERweb) and the county's nightly Buildings.csv.
Hidden Water Feature: proven October 16, 2026 on 18 test cases (12 in the corridor, 6 elsewhere in Colorado), precision 1.00, recall 1.00, against USGS NHD waterbodies, flowlines and points, and the USFWS National Wetlands Inventory (public ArcGIS, no key).
Water Rights Well: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USGS NWIS site service, groundwater sites (public, no key).
Census Flip Tract: proven October 16, 2026 on 24 test cases (1 in the corridor, 9 elsewhere in Colorado, 8 in Virginia, 6 elsewhere in the U.S.), precision 1.00, recall 1.00, against U.S. Census Bureau ACS 5-year summary files 2021 and 2024 (www2.census.gov, no key).
Micro Trend Radar: proven October 16, 2026 on 19 test cases (19 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales and buildings (the adapter from the public map services, the label from the county's nightly public files).
Boom Town Follower: proven October 18, 2026 on 17 test cases (17 elsewhere in Colorado), precision 1.00, recall 1.00, against U.S. Census Bureau Vintage 2024 county population estimates, co-est2024-alldata.csv (www2.census.gov, no key).
Income Stack Score: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Six side-income layers, each the label of a signal that passed acceptance.
Cost To Activate Gap: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor nightly Buildings.csv.
Best Fit Buyer: proven October 18, 2026 on 26 test cases (26 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor nightly files (owners classed estate or business, values, sales), the stalled_permit_distress, code_violation_x_tall_grass and Acreage labels (county permits, code cases and parcels; public, no key).
Keep One Wall Tax: proven October 18, 2026 on 24 test cases (12 in the corridor, 12 in Virginia), precision 1.00, recall 1.00, against Va. Code 58.1-3220 and the ordinances of Chesterfield County, Henrico County and Harrisonburg (official pages, read 2026-10-18); Colorado law (no such exemption).
House Hack Fit: proven October 18, 2026 on 24 test cases (8 in the corridor, 16 elsewhere in Colorado), precision 1.00, recall 1.00, against U.S. Census Bureau ACS 2020-2024 5-year summary file, table B25064 (median gross rent), per tract and county (www2.census.gov, no key).
Per Parcel Real Costs: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor, estimated taxes by taxing authority (public ArcGIS, no key).
On Market Deal Score: proven October 18, 2026 on 37 test cases (37 in the corridor), precision 1.00, recall 1.00, against Zillow alerts in the shinnslist inbox matched to county parcels; the parcel's county actual value (parcel store) rolled with the FHFA House Price Index (fhfa.gov) and comparable recorded sales (parcel store).
Whole Region Rank: proven October 16, 2026 on 24 test cases (24 in Virginia), precision 1.00, recall 1.00, against VGIN statewide parcels (Virginia, public ArcGIS, no key).
Zoning Arbitrage: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Corridor zoning maps in Boulder County, county parcels and the nightly Buildings.csv.
Mislisted Sqft: proven October 18, 2026 on 117 test cases (117 in the corridor), precision 1.00, recall 1.00, against listings inbox card facts against Boulder CAMA, Larimer Assessor Public Data Center and Jefferson jMap building records (public, no key).
Improvement Zone Between Hnw: proven October 18, 2026 on 24 test cases (13 in the corridor, 11 elsewhere in Colorado), precision 1.00, recall 1.00, against U.S. Census Bureau ACS 2020-2024 5-year summary file, table B19013 (median household income), per tract and county (www2.census.gov, no key).
No Airbnb License Acreage: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County short-term and vacation rental license register, assessor parcels and zoning map (public map services, no key).
Sprinter Van Summer Camp: proven October 16, 2026 on 24 test cases (11 in the corridor, 13 in Virginia), precision 1.00, recall 1.00, against USDA Forest Service recreation opportunities and The National Map campgrounds (public map services, no key).
Mixed Zoning Live Work: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Corridor zoning maps and the towns' and counties' use tables.
Alphaearth Core: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against AlphaEarth Foundations annual satellite embeddings (Google and Google DeepMind; public copy on Source Cooperative, CC-BY 4.0), 2017 against 2025 at the parcel point's 10 m pixel, Boulder County corridor parcels.
Code Violation X Tall Grass: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Land Use, code enforcement cases (Accela, public ArcGIS layer, no key).
Tax Delinquency Early Warning Formula: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Larimer County Treasurer, 2026 tax lien sale list (shinnslist's stored copy) and the Assessor's tax parcel layer.
Auction Underbidder Join: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales and owner files (nightly CSVs, no key).
Life Event Distress: proven October 18, 2026 on 18 test cases (1 in the corridor, 17 elsewhere in Colorado), precision 1.00, recall 1.00, against U.S. Census Bureau ACS 2020-2024 5-year summary file, table B23025 (employment status), per county (www2.census.gov, no key).
Foreclosure Surplus Join: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales and values files (nightly CSVs, no key).
Unpermitted To Sweat Equity: proven October 18, 2026 on 23 test cases (12 elsewhere in Colorado, 1 in Virginia, 10 elsewhere in the U.S.), precision 1.00, recall 0.91, against Socrata Open Data catalog API (api.us.socrata.com/api/catalog/v1, no key).
Tax Delinquency Early Warning: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Larimer County Treasurer's 2026 tax lien sale advertising list of delinquent real property (public Excel file, refreshed nightly until the 2026-11-19 sale), joined through Larimer County's tax parcels layer.
OpenFEMA Flood Buyout: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against FEMA National Flood Hazard Layer, flood hazard zones (public ArcGIS, no key).
Burn Scar / Fire Channel: proven October 16, 2026 on 24 test cases (12 in the corridor, 2 elsewhere in Colorado, 10 elsewhere in the U.S.), precision 1.00, recall 1.00, against NIFC WFIGS Interagency Perimeters Current (public ArcGIS, no key).
AMI Smart Meter Vacancy: proven October 18, 2026 on 20 test cases (8 in the corridor, 12 elsewhere in Colorado), precision 1.00, recall 1.00, against U.S. Census Bureau ACS 2020-2024 5-year summary files, tables B25002 (occupancy status) and B25004 (vacancy status), per tract (www2.census.gov, no key).
Water X Distress Matrix: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against FEMA National Flood Hazard Layer, flood hazard zones (public ArcGIS, no key) crossed with Larimer County Treasurer's 2026 tax lien sale list (shinnslist's stored copy) through Larimer County's tax parcel layer.
Foreclosure Nod Join: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales file (every recorded document per account; nightly CSV, no key).
Probate Preprobate X Parcel: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor, parcel owners map layer (public ArcGIS, no key).
Absentee Entity High Equity: proven October 16, 2026 on 48 test cases (48 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor's parcel owners, sales and values map layers, and Larimer County's tax parcels layer with its Assessor's public value and sales files (public, no key).
Estate Sales: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales file (nightly CSV, no key).
Dumpster Out Front: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against City of Boulder Construction Permits (city open data).
Unkept Property: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 0.92, against USDA NAIP 4-band imagery (Microsoft Planetary Computer.
Retirement Home Visitors: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against CMS Care Compare nursing homes (Provider Information, data.cms.gov).
Fb Marketplace Downsizing: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor (accounts, buildings, sales; public map services).
Three Percent Mortgage Buyers List: proven October 18, 2026 on 24 test cases (24 elsewhere in Colorado), precision 1.00, recall 1.00, against U.S. Census Bureau ACS 2020-2024 5-year summary file, table B25081 (mortgage status), per county and the nation (www2.census.gov, no key).
Confidence Percent: proven October 18, 2026 on 10 test cases (10 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up hunt rows scored by the confidence engine (lib/confidence-score.mjs) against its frozen tier counts (lib/confidence-tiers.json, snapshot 2026-10-16).
Add Dataset Extensible: proven October 18, 2026 on 12 test cases (12 elsewhere in the U.S.), precision 1.00, recall 1.00, against own data.
Wrong Path Killswitch: proven October 18, 2026 on 8 test cases (8 elsewhere in the U.S.), precision 1.00, recall 1.00, against own code.
Assumable / Seller Finance Flag: proven October 18, 2026 on 24 test cases (2 in the corridor, 22 elsewhere in Colorado), precision 1.00, recall 1.00, against CFPB / FFIEC HMDA Data Browser API, /view/aggregations, 2020-2022 home-purchase loan application records per county (ffiec.cfpb.gov, public domain, no key).
Rent Back Timeshare: proven October 18, 2026 on 48 test cases (48 in the corridor), precision 1.00, recall 1.00, against Summit, Routt and Archuleta County Assessors' own parcel records (the three Colorado counties whose free record classes timeshare units).
STR Attached Home Premium: proven October 18, 2026 on 40 test cases (40 in the corridor), precision 1.00, recall 1.00, against STR license registers (City of Boulder, Boulder/Summit/Larimer/Clear Creek counties, Town of Avon; public map services, no key), county assessor building records, ACS 2020-2024 B25024 summary file, TIGERweb places.
Dscr Projected Income List: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against HUD small-area rents, the county's nightly Buildings.csv and Values.csv, its tax table, and Freddie Mac's weekly 30-year average.
Middle Man Lending: proven October 18, 2026 on 20 test cases (1 in the corridor, 19 elsewhere in Colorado), precision 1.00, recall 1.00, against CFPB / FFIEC HMDA Data Browser API, /view/aggregations (ffiec.cfpb.gov, no key).
Lead At Close Commission: proven October 18, 2026 on 10 test cases (10 elsewhere in the U.S.), precision 1.00, recall 1.00, against checkout attempts against the real handler with a recorded (never reached) Stripe endpoint.
Unique Home Lender Match: proven October 18, 2026 on 24 test cases (1 in the corridor, 23 elsewhere in Colorado), precision 1.00, recall 1.00, against CFPB / FFIEC HMDA Data Browser API, /view/aggregations (ffiec.cfpb.gov, no key).
Unique Home Insurance Match: proven October 18, 2026 on 24 test cases (2 in the corridor, 22 elsewhere in Colorado), precision 1.00, recall 1.00, against U.S. Census Bureau ACS 2020-2024 5-year summary file, table B25032 (tenure by units in structure), per county (www2.census.gov, no key).
Paperwork Sherpa: proven October 18, 2026 on 11 test cases (11 elsewhere in the U.S.), precision 1.00, recall 1.00, against the library’s Commission form families and version checks recorded through the real handler (made-up checks for the test).
Project 100 Zip Leads: proven October 18, 2026 on 24 test cases (24 elsewhere in Colorado), precision 1.00, recall 1.00, against CFPB / FFIEC HMDA Data Browser API, /view/filers per county (ffiec.cfpb.gov, no key).
Ispeedtolead Clone Shape: proven October 18, 2026 on 15 test cases (15 elsewhere in the U.S.), precision 1.00, recall 1.00, against own data.
Land Renewable Lease Loan: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against EIA U.S. Energy Atlas Electric Substations (ArcGIS FeatureServer, public, no key), read in bulk by 0.1-degree tile.
Wholetail Private Master: proven October 18, 2026 on 13 test cases (13 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up parcels worked on the deal sheet (lib/deal-sheet.js applyDeal.
Off Market Business Inventory: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Colorado Secretary of State business entities (data.colorado.gov), Boulder County assessor parcels and sales.
Live Stream Whats Working: proven October 18, 2026 on 8 test cases (8 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up truth-table rows, and the committed truth table for two real signals.
User Friendly Wholesale App: proven October 18, 2026 on 7 test cases (7 elsewhere in the U.S.), precision 1.00, recall 1.00, against one made-up Boulder parcel per case in a fresh database.
Pay Per Lead List: proven October 18, 2026 on 11 test cases (11 elsewhere in the U.S.), precision 1.00, recall 1.00, against checkout attempts against the real handler with a recorded (never reached) Stripe endpoint.
Model Access Tokens: proven October 18, 2026 on 11 test cases (11 elsewhere in the U.S.), precision 1.00, recall 1.00, against own data.
Jmu Compound Hunt: proven October 18, 2026 on 10 test cases (10 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up Virginia and Colorado parcels with stored outline acres, water kinds and proven tags.
Dream Inventory Camps: proven October 18, 2026 on 9 test cases (9 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up Colorado and Virginia parcels with stored compound kinds (camp), water kinds and proven tags.
Auto Dossiers: proven October 18, 2026 on 18 test cases (18 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up test parcels in a fresh database.
Outreach Prep Deal Desk: proven October 18, 2026 on 12 test cases (12 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up parcels with test owners and mailing addresses, suppressions (migration 0011) and owner removals.
Space To Earth Zoom: proven October 18, 2026 on 9 test cases (9 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up parcels at real corridor and non-corridor points in a fresh database; assets/places.json for the town names.
Epic Viz Target Mix: proven October 18, 2026 on 8 test cases (8 elsewhere in the U.S.), precision 1.00, recall 1.00, against six made-up Colorado parcels in a fresh database.
Click Pick 50Plus Lists: proven October 18, 2026 on 114 test cases (114 elsewhere in the U.S.), precision 1.00, recall 1.00, against own data.
Watch Mode Alerts: proven October 18, 2026 on 20 test cases (20 elsewhere in the U.S.), precision 1.00, recall 1.00, against own data.
MAP Color By Edge: proven October 18, 2026 on 7 test cases (7 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up frozen scores in the hunt-scores.json shape ({ parcels.
Freshness Repull: proven October 18, 2026 on 7 test cases (7 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up registry entries, a stub count service (the live path is the curl count query).
Public Prediction Ledger: proven October 18, 2026 on 6 test cases (6 elsewhere in the U.S.), precision 1.00, recall 1.00, against tests/fixtures/confidence-universe-tiny.ndjson (ten made-up parcels).
Verified Lead Ledger: proven October 18, 2026 on 6 test cases (6 elsewhere in the U.S.), precision 1.00, recall 1.00, against tests/fixtures/confidence-universe-tiny.ndjson frozen on 2026-01-02, one made-up listing per case.
Interactive Deal Room: proven October 18, 2026 on 10 test cases (10 elsewhere in the U.S.), precision 1.00, recall 1.00, against two made-up parcels and a stored check in a fresh database; a made-up owner password that exists only in the run.
Github Opensource Signal Mine: proven October 18, 2026 on 6 test cases (6 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up search answers in GitHub's shape; the live run of 2026-10-18 is data/signal-proposals/github-2026-10-18.json.
Reso MLS Tailwind: proven October 18, 2026 on 7 test cases (7 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up alert emails in Zillow's card shape.
Proprietary Models Outputs Only: proven October 18, 2026 on 10 test cases (10 elsewhere in the U.S.), precision 1.00, recall 1.00, against two made-up parcels in a fresh database; the engine's own freeze summary and weights table as the known-internal controls.
Signal Gating Ledger: proven October 18, 2026 on 18 test cases (18 elsewhere in the U.S.), precision 1.00, recall 1.00, against the committed signal truth table and the agent catalog that reads it.
Cabin: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 0.92, against Boulder County Assessor appraisal records (public map services, no key).
Farm: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USDA NRCS SSURGO through Soil Data Access (public, no key).
Off Grid: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County onsite wastewater (septic) system records (public ArcGIS, no key).
Prepped: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Colorado DWR well permits and USFS wildfire hazard (their proven signals' labels) and FEMA's flood hazard zones (NFHL).
Horse / Equestrian: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor appraisal records (public map services, no key).
Waterfall: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USGS NHD points, FCode 48700 waterfall (public ArcGIS, no key).
River Frontage, Rights & Flood Risk: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USGS NHD large-scale flowlines (public ArcGIS, no key).
Greenhouse / Attached Greenhouse Houses: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor appraisal records (public map services, no key).
Acreage: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 0.92, against Boulder County Assessor parcel records (public map service, no key).
Hunting: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USGS PAD-US public access layer (fee land, open access), ArcGIS REST, no key.
Ranching: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USDA NRCS SSURGO through Soil Data Access (public, no key).
Old American Mansions / Architectural Styles: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor appraisal records (public map services, no key).
Investor Specials: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County code cases and permits (builder 1's proven labels) and the Assessor's nightly Values.csv, Buildings.csv and Sales.csv.
Barndominiums: proven October 16, 2026 on 13 test cases (13 in the corridor), precision 1.00, recall 1.00, against Boulder County land-use and building records (Accela) and zoning map (public map services, no key).
Cities In Other Countries + Translation: proven October 18, 2026 on 7 test cases (7 elsewhere in the U.S.), precision 1.00, recall 1.00, against HM Land Registry Price Paid Data (City of London, 2026 sales; excerpt in tests/fixtures/intl-ppd-city-of-london.csv) and INSPIRE Index Polygons (City of London), Open Government Licence; postcode centres from postcodes.io (ONS, OGL).
Paranormal Listings In Areas With History: proven October 16, 2026 on 24 test cases (3 in the corridor, 21 in Virginia), precision 1.00, recall 1.00, against NPS National Register of Historic Places, USGS The National Map (cemeteries, historic sites, GNIS historical places) and Boulder County historic sites (public map services, no key).
Bigfoot Listings: proven October 16, 2026 on 24 test cases (1 in the corridor, 23 elsewhere in Colorado), precision 1.00, recall 1.00, against Bigfoot Field Researchers Organization county report pages (bfro.net) and Census TIGERweb counties.
Earthships: proven October 16, 2026 on 16 test cases (16 in the corridor), precision 1.00, recall 1.00, against Boulder County land-use and building records (Accela) and zoning map (public map services, no key).
Hobbit Homes: proven October 16, 2026 on 15 test cases (15 in the corridor), precision 1.00, recall 1.00, against Boulder County land-use and building records (Accela) and zoning map (public map services, no key).
Recording Studios: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County land-use and building records (Accela) and zoning map (public map services, no key).
Up-And-Coming Vacation Properties: proven October 18, 2026 on 21 test cases (21 elsewhere in Colorado), precision 1.00, recall 1.00, against U.S. Census Bureau ACS 2020-2024 5-year summary file, table B07001 (geographical mobility in the past year), per county and the nation (www2.census.gov, no key).
Interior Design Styles: proven October 18, 2026 on 123 test cases (123 in the corridor), precision 1.00, recall 1.00, against listings inbox card facts against Boulder CAMA, Larimer Assessor Public Data Center and Jefferson jMap building records (public, no key) (the building architectural style in the record, not interior photos).
Multi Source Re Aggregator: proven October 18, 2026 on 5 test cases (5 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up parcels and stored check answers in a fresh database.
Rental Apartment Junk Filter: proven October 16, 2026 on 24 test cases (4 in the corridor, 6 elsewhere in Colorado, 14 in Virginia), precision 1.00, recall 1.00, against HUD Multifamily Properties - Assisted (public ArcGIS, no key).
Click Pick List Builder Api: proven October 18, 2026 on 17 test cases (17 elsewhere in the U.S.), precision 1.00, recall 1.00, against own data.
STR Analysis Site Pattern: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against City of Boulder rental licenses (short-term, Issued) and Boulder County's short-term and vacation rental register (active records).
House For Equity Swap Listings: proven October 18, 2026 on 10 test cases (10 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up sign-ups (example.test addresses) on a fresh board.
Public Land Access Verify: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USDA Forest Service National Forest boundaries and BLM surface-management layer (public ArcGIS, no key).
Self Healing Scraper Mcp: proven October 18, 2026 on 6 test cases (6 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up registry entry and recorded fallbacks, a stub count service.
Openclaw Wrapper Mrr Pattern: proven October 18, 2026 on 7 test cases (7 elsewhere in the U.S.), precision 1.00, recall 1.00, against one made-up parcel in a fresh database; the committed llms.txt. Listing the connection in agent directories waits on James and is not done here..
Vacant Nod Zombie: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales and owner files (nightly CSVs, no key).
Code Lien Foreclosure: proven October 18, 2026 on 16 test cases (16 in the corridor), precision 1.00, recall 1.00, against Boulder County Land Use cases (Accela layer, read whole) and the Assessor sales file.
Early Change Motivation Score: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Four signals proven on corridor parcels, read live by the composite.
Fresh Verified Distress Stack: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against The same proven distress legs, each dated by what it read (the Boulder files' CreatedDate, the Larimer files' RUNDATE, the stored tax list's as-of date, live layers today).
Parcel Buildability Score: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USDA NRCS soil survey through Soil Data Access (the 'Dwellings without basements' rating of the soil component covering most of the map unit under the point) and FEMA's flood hazard zones (public, no key).
Stalled Permit Distress: proven October 16, 2026 on 48 test cases (48 in the corridor), precision 1.00, recall 1.00, against Boulder County building permits (Accela, public ArcGIS layer, no key) and the City of Boulder construction permits (public ArcGIS layer, no key).
Heir Property Fractional: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales and owner files (nightly CSVs, no key).
Fire Casualty Permit: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County fire-casualty records (Accela Four Mile Fire and Marshall Fire records; the Marshall Fire parcel layer; public ArcGIS, no key).
Investor Concentration Index: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor's parcel owners layer (public ArcGIS, no key).
Vacant Land Owner Distress Join: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales and owner files (nightly CSVs, no key).
Holder Mix Map: proven October 18, 2026 on 24 test cases (11 in the corridor, 13 elsewhere in Colorado), precision 1.00, recall 1.00, against U.S. Census Bureau ACS 2020-2024 5-year summary file, table B25003 (tenure), per tract and county (www2.census.gov, no key).
Property Condition Risk Score: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor nightly building file (Buildings.csv).
Listing Truth Gap: proven October 18, 2026 on 117 test cases (117 in the corridor), precision 1.00, recall 1.00, against listings inbox card facts against Boulder CAMA, Larimer Assessor Public Data Center and Jefferson jMap building records (public, no key).
Institutional Owner Entity Resolution: proven October 18, 2026 on 38 test cases (38 elsewhere in the U.S.), precision 1.00, recall 1.00, against SEC EDGAR company search (www.sec.gov/cgi-bin/browse-edgar, action=getcompany, no key).
Tenant Turnover Maintenance Risk: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against U.S. Census Bureau ACS 5-year table B25038 (Census Reporter, every tract of Boulder and Larimer counties) and Census 2020 tracts (TIGERweb Tracts_Blocks).
Hoa Superpriority Lien: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales and owner files (nightly CSVs, no key).
Commercial Listing Staleness Score: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Colorado Secretary of State business entities (data.colorado.gov) and Boulder County assessor parcels.
Gold Stream Prospecting: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USGS Mineral Resources Data System (MRDS) WFS, layer mrds-high (public, no key).
Waterfall Build Site: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USGS NHD points, FCode 48700 waterfall (public ArcGIS, no key).
List Saturation Index: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against shinnslist listings inbox (Zillow saved-search alerts to listings@shinnslist.com, read-only).
Seller Trust Score: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales file (nightly CSV, no key).
Rehab Cost Estimator: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor nightly Buildings.csv and Values.csv and the City of Boulder's declared costs for single-family alterations.
Data Center Land Siting: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against HIFLD electric substations (public ArcGIS mirror, no key).
Wildfire Insurance Mitigation: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County wildfire zones under its building code (Section R390.2, adopted July 1, 2026; public ArcGIS, no key).
Derelict Utilities Knockdown: proven October 18, 2026 on 24 test cases (7 in the corridor, 17 elsewhere in Colorado), precision 1.00, recall 1.00, against U.S. Census Bureau ACS 2020-2024 5-year summary files, tables B25004 (vacancy status, line 8 "other vacant") and B25002 (occupancy status, line 1 total housing units), per tract (www2.census.gov, no key).
Seasonal Water Temporal: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USGS NHD large-scale flowlines (public ArcGIS, no key).
Water Diversion Pond Feasibility: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USGS National Hydrography Dataset, large scale (hydro.nationalmap.gov, public, no key).
Camp Lodge Cabin Conversion: proven October 16, 2026 on 24 test cases (16 in the corridor, 8 in Virginia), precision 1.00, recall 1.00, against OpenStreetMap (tourism=camp_site, chalet, alpine_hut; leisure=resort) via Photon category search and Nominatim outlines (no key; © OpenStreetMap contributors, ODbL).
Bedroom Add Arbitrage: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales, buildings and appraisal neighborhoods (the adapter from the public map services with warranty deeds standing in; the label from the nightly Sales.csv sales the Assessor qualified, Buildings.csv, and the parcel layer's NbhdCode read whole).
Lot Split Subdivision Potential: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Corridor zoning maps (Boulder County and its towns, Estes Park, Larimer County, Erie) and county parcels.
Adu Duplex Eligibility: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Corridor zoning maps, county parcels, Boulder County's nightly Buildings.csv, FEMA flood zones, the towns' and counties' codes and Colorado HB24-1152.
STR Lease Arbitrage Absentee: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County parcel owners, zoning, parcels, buildings and short-term and vacation rental licenses (public ArcGIS, no key) and the Census county subdivisions, under Boulder County Ordinance 2023-02.
Military Assumable Loan: proven October 18, 2026 on 24 test cases (12 in the corridor, 12 elsewhere in Colorado), precision 1.00, recall 1.00, against U.S. DoD installation boundaries (FY2024 MIRTA), USDOT/BTS NTAD Military Bases FeatureServer (public, no key), read in bulk by 0.1-degree tile.
Hyperlocal Catalyst Monitor: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County land-use docket and building permits (Accela) and zoning map (public map services, no key).
Distress Stack Conviction Score: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against The corridor's proven distress signals (Boulder County.
Offmarket Probability Engine: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against The proven distress legs and the county's last sale of $10,000 or more against today's value (Boulder Sales.csv and Values.csv; Larimer sales and value detail files).
Assumable Equity Arbitrage: proven October 18, 2026 on 24 test cases (2 in the corridor, 22 elsewhere in Colorado), precision 1.00, recall 1.00, against CFPB / FFIEC HMDA Data Browser API, /view/aggregations, 2020-2022 home-purchase loan application records per county (ffiec.cfpb.gov, public domain, no key).
Teardown Option Value: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor, land and building values (public ArcGIS, no key).
Water Value Stack: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USGS National Hydrography Dataset, large scale (hydro.nationalmap.gov, public, no key).
STR Regulatory Arbitrage Map: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Corridor STR ordinances (Boulder, Lyons, Longmont, Lafayette; Boulder County Ordinance 2023-02), Boulder County zoning and short-term and vacation rental licenses, Census county divisions, and the hunt's STR investor towns.
Insurance Mitigation Value: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against FEMA National Flood Hazard Layer, flood hazard zones (public ArcGIS, no key).
Lien Stack Priority: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales file and Land Use cases.
Heir Fractional Assemblage: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor owner file (nightly CSV, no key).
Catalyst Front Run: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County land-use docket (Accela) and zoning map (public map services, no key).
Bedroom Adu Value Uplift: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against The bedroom_add_arbitrage label (Boulder County nightly sales and buildings) and the adu_duplex_eligibility label (corridor zoning maps, county parcels and buildings, FEMA flood zones, the codes).
Adverse Possession Ripening Join: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales and owner files (nightly CSVs, no key).
Motel OPM Guaranteed Occupancy: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor nightly Buildings.csv (class 2215 buildings designed as hotels, lodging or inns) joined to Account_Parcels.csv.
Grant Stacked Land Yield: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USGS PAD-US 4.1 easements managed by NRCS, Census TIGERweb counties and USAspending.gov county obligations for EQIP, ACEP, CRP and CSP (public, no key).
Absentee STR Conversion Funnel: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor owner and sales files (nightly CSVs, no key).
Tired Landlord Composite: proven October 16, 2026 on 48 test cases (48 in the corridor), precision 1.00, recall 1.00, against Boulder County parcel owners and sales and its code cases and permits (Accela), and Larimer County's tax parcels layer, its Assessor's public sales file and its Treasurer's tax lien sale list, public, no key.
Squatter Occupancy Signal: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder Police Department offenses (City of Boulder open data).
Tired Landlord to Seller: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against City of Boulder Rental Housing Licenses (city open data) and Boulder County assessor sales.
Geotagged Photo Distress Mine: proven October 18, 2026 on 24 test cases (16 in the corridor, 8 elsewhere in Colorado), precision 1.00, recall 1.00, against Wikimedia Commons category tree.
Neglect Imagery Timeseries: proven October 18, 2026 on 22 test cases (22 in the corridor), precision 1.00, recall 0.90, against USDA NAIP 4-band imagery (Microsoft Planetary Computer.
Yard To Storage Conversion: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor land and values files (nightly CSVs, no key).
Assumable Pool Network: proven October 18, 2026 on 11 test cases (11 elsewhere in the U.S.), precision 1.00, recall 1.00, against made-up parcels and one test buyer sign-up (counties Adams, Weld, Boulder; up to $500,000); FHA/VA shares from the stored HMDA copy (2026-10-16).
Search Intent Distress Trend: proven October 16, 2026 on 24 test cases (13 in the corridor, 11 elsewhere in Colorado), precision 1.00, recall 1.00, against CFPB Consumer Complaint Database (search API).
Mechanic Lien Distress: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Land Use cases (Accela layer, read whole).
Parcel Insurance Cost Insurability: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against FEMA NFHL flood hazard zones (public ArcGIS) and USDA Forest Service Wildfire Hazard Potential 2023, classified (federal GeoPlatform image service), both no key.
Deed Fraud Watch: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales and owner files (nightly CSVs, no key).
Marketing Compliance Risk: proven October 18, 2026 on 22 test cases (1 elsewhere in Colorado, 1 in Virginia, 20 elsewhere in the U.S.), precision 1.00, recall 1.00, against FCC Consumer Complaints Data, Unwanted Calls (opendata.fcc.gov Socrata vakf-fz8e, no key), complaints by state over the trailing 12 months; Census Bureau Vintage 2024 state population estimates (NST-EST2024-ALLDATA, www2.census.gov).
Appraisal Gap Risk: proven October 18, 2026 on 40 test cases (40 in the corridor), precision 1.00, recall 1.00, against Zillow alerts in the shinnslist inbox matched to county parcels; the highest of the county actual value (parcel store) rolled with the FHFA House Price Index and the comparable-sales estimate (parcel store).
Post Purchase Tax Reassessment: proven October 18, 2026 on 40 test cases (40 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales, values and owner files (nightly CSVs, no key).
Self Storage Retiring Owner: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor buildings and sales files (nightly CSVs, no key).
Commercial Distress Vacancy Signal: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Colorado Secretary of State business entities (data.colorado.gov) and Boulder County assessor parcels.
Rent Control Regulatory Risk: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Colorado C.R.S. 38-12-301 (rent control preempted) and the corridor's long-term rental licensing ordinances (City of Boulder, Denver, Fort Collins; Boulder County licenses none), placed with Boulder County's zoning map, Larimer County's parcel layer and the parcel's county.
Insurance Claim Denial: proven October 16, 2026 on 22 test cases (4 in the corridor, 5 elsewhere in Colorado, 13 in Virginia), precision 1.00, recall 1.00, against OpenFEMA NFIP redacted claims v3 and the Census geocoder (federal, no key).
Rent Payment Reliability: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against HUD Multifamily Assisted Properties (HUD map service).
Eviction Timeline Risk: proven October 18, 2026 on 24 test cases (8 in the corridor, 16 elsewhere in Colorado), precision 1.00, recall 1.00, against U.S. Census Bureau ACS 2020-2024 5-year summary file, table B25071 (median gross rent as a percentage of household income), per tract (www2.census.gov, no key).
Damage Protection Gap: proven October 18, 2026 on 24 test cases (21 in the corridor, 3 elsewhere in Colorado), precision 1.00, recall 1.00, against FEMA NFHL flood hazard zones (public ArcGIS), Census TIGERweb tracts and OpenFEMA NFIP Redacted Policies v3, all public and no key.
Platform Support Failure: proven October 16, 2026 on 23 test cases (14 in the corridor, 9 elsewhere in Colorado), precision 1.00, recall 1.00, against CFPB Consumer Complaint Database (search API).
Well Depth Drill Cost: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against USGS NWIS site service, groundwater sites with a reported depth (public, no key).
Ai Title Abstract: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales file (nightly CSV, no key).
Condo Assessment Distress Join: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales and owner files (nightly CSVs, no key).
Mineral Royalty Suspense Join: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales file (nightly CSV, no key).
Agent Query Demand Join: proven October 18, 2026 on 16 test cases (16 elsewhere in the U.S.), precision 1.00, recall 1.00, against own data.
Aggregate Reserve Distress Join: proven October 16, 2026 on 24 test cases (12 in the corridor, 12 in Virginia), precision 1.00, recall 1.00, against USGS Mineral Resources Data System (WFS mrds-high, no key).
Air Rights Sending Site Join: proven October 18, 2026 on 24 test cases (22 in the corridor, 2 elsewhere in Colorado), precision 1.00, recall 1.00, against NPS National Register of Historic Places locations (points and boundaries, public domain).
Airport Noise Contour Join: proven October 16, 2026 on 24 test cases (24 in Virginia), precision 1.00, recall 1.00, against FAA airport master records (5010) via BTS NTAD Aviation Facilities (public ArcGIS, no key).
Assumable Mortgage Under 4pct: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor sales (the adapter from the public map service, the label from the nightly Sales.csv) and Freddie Mac's Primary Mortgage Market Survey.
Bank Branch Closure Join: proven October 16, 2026 on 24 test cases (6 in the corridor, 18 in Virginia), precision 1.00, recall 1.00, against FDIC BankFind institution history, branch closings (change code 721; public API, no key).
BEAD Site Join: proven October 16, 2026 on 24 test cases (6 in the corridor, 18 elsewhere in Colorado), precision 1.00, recall 1.00, against Colorado Broadband Office BEAD final proposal locations and USGS 3DEP elevation (public, no key).
BESS Site Join: proven October 16, 2026 on 24 test cases (5 in the corridor, 19 in Virginia), precision 1.00, recall 1.00, against HIFLD electric substations (public copy on ArcGIS Online, no key).
Billboard Frontage Screen Join: proven October 16, 2026 on 24 test cases (4 in the corridor, 20 in Virginia), precision 1.00, recall 1.00, against FHWA National Highway System with AADT and FHWA National Scenic Byways via NTAD (public ArcGIS, no key).
Bowling Alley Distress Join: proven October 16, 2026 on 24 test cases (10 elsewhere in Colorado, 14 in Virginia), precision 1.00, recall 1.00, against OpenStreetMap (leisure.
Brightfield Solar Join: proven October 16, 2026 on 24 test cases (6 in the corridor, 18 in Virginia), precision 1.00, recall 1.00, against EPA RE-Powering Mapper sites screened for solar (public ArcGIS, no key).
Brownfield Credit Join: proven October 16, 2026 on 24 test cases (8 in the corridor, 16 in Virginia), precision 1.00, recall 1.00, against EPA Facility Registry Service brownfields properties (ACRES; public ArcGIS, no key).
Built 100+ Years Ago: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor appraisal records (public map services, no key).
Buyout Program Enrollment Join: proven October 16, 2026 on 24 test cases (8 in the corridor, 16 in Virginia), precision 1.00, recall 1.00, against FEMA NFHL flood zones, OpenFEMA HMA acquisition subapplications and Census TIGERweb counties (public, no key).
Camp Retreat Conversion Join: proven October 16, 2026 on 24 test cases (6 in the corridor, 1 elsewhere in Colorado, 17 in Virginia), precision 1.00, recall 1.00, against OpenStreetMap (leisure.
Closed Facility Conversion Join: proven October 16, 2026 on 24 test cases (5 in the corridor, 1 elsewhere in Colorado, 11 in Virginia, 7 elsewhere in the U.S.), precision 1.00, recall 1.00, against HIFLD hospitals and nursing homes (FEMA preparedness copies, public ArcGIS, no key).
Coal-to-Nuclear Siting Join: proven October 16, 2026 on 24 test cases (3 in the corridor, 1 elsewhere in Colorado, 12 in Virginia, 8 elsewhere in the U.S.), precision 1.00, recall 1.00, against EIA power plants, EIA-860/860M (public ArcGIS copy, no key).
Conservation Burial Grounds Join: proven October 16, 2026 on 24 test cases (3 in the corridor, 21 in Virginia), precision 1.00, recall 1.00, against OpenStreetMap cemeteries via Photon category search and Nominatim outlines (© OpenStreetMap contributors, ODbL) and FEMA NFHL flood zones (public, no key).
Conservation Easement Credit Join: proven October 16, 2026 on 24 test cases (7 in the corridor, 17 in Virginia), precision 1.00, recall 1.00, against USGS PAD-US 4.1 easements (public ArcGIS, no key).
Correctional Closure Conversion Join: proven October 16, 2026 on 24 test cases (15 in Virginia, 9 elsewhere in the U.S.), precision 1.00, recall 1.00, against HIFLD prison boundaries (FEMA preparedness copy, public ArcGIS, no key).
Dealership Distress Join: proven October 16, 2026 on 24 test cases (1 in the corridor, 23 in Virginia), precision 1.00, recall 1.00, against OpenStreetMap (shop.
Dormant Mineral Abandonment Join: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor appraisal records (mineral accounts and sales; public map services, no key).
Drive-In Closure Join: proven October 16, 2026 on 24 test cases (3 in the corridor, 1 elsewhere in Colorado, 12 in Virginia, 8 elsewhere in the U.S.), precision 1.00, recall 1.00, against OpenStreetMap (amenity.
EGS Geothermal Site Join: proven October 18, 2026 on 24 test cases (12 in the corridor, 12 elsewhere in Colorado), precision 1.00, recall 0.92, against IHFC Global Heat Flow Database, release 2024 (GFZ Data Services, CC BY 4.0).
Exchange Window Join: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor appraisal records (sales and parcel tables; public map services, no key).
Farm or Ranch: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor appraisal records (public map services, no key).
Federal Grant Round Join: proven October 16, 2026 on 24 test cases (9 in the corridor, 15 in Virginia), precision 1.00, recall 1.00, against FEMA NFHL flood zones, EPA FRS brownfields (ACRES) and the Grants.gov search API (public, no key).
Forest Carbon Enrollment Join: proven October 16, 2026 on 24 test cases (6 in the corridor, 18 in Virginia), precision 1.00, recall 1.00, against USGS/USFS NLCD 2021 land cover and tree canopy (MRLC map service) and USGS PAD-US 4 fee lands (public, no key).
Geothermal Spring Parcel Join: proven October 16, 2026 on 24 test cases (4 in the corridor, 9 elsewhere in Colorado, 11 in Virginia), precision 1.00, recall 1.00, against NOAA NCEI Thermal Springs in the United States (public ArcGIS, no key).
Golf Course Conversion Join: proven October 16, 2026 on 24 test cases (2 in the corridor, 22 in Virginia), precision 1.00, recall 1.00, against OpenStreetMap (leisure.
Grazing Allotment AUM Join: proven October 16, 2026 on 24 test cases (12 in the corridor, 12 in Virginia), precision 1.00, recall 1.00, against BLM national grazing allotments and Forest Service active range allotments (public ArcGIS, no key).
Historic Credit Eligibility Join: proven October 16, 2026 on 24 test cases (24 in Virginia), precision 1.00, recall 1.00, against National Park Service National Register of Historic Places boundaries (public ArcGIS, no key).
Historic Farm, Mill or Plantation: proven October 18, 2026 on 24 test cases (21 in the corridor, 3 elsewhere in Colorado), precision 1.00, recall 1.00, against National Park Service National Register of Historic Places boundaries (public ArcGIS, no key).
Hunting Lease Income Join: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Colorado Parks and Wildlife Species Activity Mapping, Elk Overall Range (public ArcGIS, no key).
Incentive Stack Eligibility Join: proven October 16, 2026 on 24 test cases (4 in the corridor, 20 in Virginia), precision 1.00, recall 1.00, against HUD Opportunity Zones, Colorado DOLA enterprise zones and VEDP Virginia enterprise zones (public ArcGIS, no key).
Insurance Non-Renewal Distress Join: proven October 16, 2026 on 24 test cases (10 in the corridor, 9 elsewhere in Colorado, 5 in Virginia), precision 1.00, recall 1.00, against USDA Forest Service Wildfire Hazard Potential 2023, classified (federal GeoPlatform image service, no key).
Landlocked Access Join: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor parcel outlines and Census TIGER public roads (public map services, no key).
LIHTC Qualified Contract Join: proven October 16, 2026 on 24 test cases (6 in the corridor, 18 in Virginia), precision 1.00, recall 1.00, against HUD Low-Income Housing Tax Credit database (public ArcGIS, no key).
Lithium Brine Rights Join: proven October 16, 2026 on 24 test cases (14 in the corridor, 10 in Virginia), precision 1.00, recall 1.00, against USGS Mineral Resources Data System (WFS mrds-high, no key).
Mall Anchor Distress Join: proven October 16, 2026 on 24 test cases (3 in the corridor, 21 in Virginia), precision 1.00, recall 1.00, against OpenStreetMap (shop.
Marina Distress Join: proven October 16, 2026 on 24 test cases (5 in the corridor, 3 elsewhere in Colorado, 16 in Virginia), precision 1.00, recall 1.00, against OpenStreetMap (leisure.
Mobile Home Park Distress Infill Join: proven October 16, 2026 on 24 test cases (5 in the corridor, 19 in Virginia), precision 1.00, recall 1.00, against Mobile home parks, HIFLD national file (FEMA preparedness copy, public ArcGIS, no key).
Mill Closure MW Join: proven October 16, 2026 on 24 test cases (1 elsewhere in Colorado, 16 in Virginia, 7 elsewhere in the U.S.), precision 1.00, recall 1.00, against EPA ECHO facility search (inactive pulp, paper and paperboard mills) and HIFLD substations (public, no key).
Miner Interconnect Distress Join: proven October 16, 2026 on 24 test cases (3 in the corridor, 21 in Virginia), precision 1.00, recall 1.00, against HIFLD electric substations (public copy on ArcGIS Online, no key).
Mitigation Bank Site Join: proven October 16, 2026 on 24 test cases (6 in the corridor, 18 in Virginia), precision 1.00, recall 1.00, against USACE RIBITS bank locations via EPA NEPAssist (public ArcGIS, no key).
Motel Conversion Distress Join: proven October 16, 2026 on 24 test cases (7 elsewhere in Colorado, 17 in Virginia), precision 1.00, recall 1.00, against OpenStreetMap (tourism.
Motel Conversion Join: proven October 16, 2026 on 24 test cases (3 in the corridor, 21 in Virginia), precision 1.00, recall 1.00, against FEMA USA Structures (ORNL footprints; public ArcGIS view, no key).
NEVI Gap Site Join: proven October 16, 2026 on 24 test cases (6 in the corridor, 1 elsewhere in Colorado, 6 in Virginia, 11 elsewhere in the U.S.), precision 1.00, recall 1.00, against FHWA Alternative Fuel Corridors (EV) and DOE AFDC fueling stations via NTAD (public ArcGIS, no key).
Nonconformity Envelope Join: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County zoning maps (county and towns), county parcels and the Assessor's building table (map services).
Office Conversion Join: proven October 16, 2026 on 24 test cases (2 in the corridor, 22 in Virginia), precision 1.00, recall 1.00, against FEMA USA Structures (ORNL footprints; public ArcGIS view, no key).
Office Conversion Screen Join: proven October 16, 2026 on 24 test cases (3 in the corridor, 21 in Virginia), precision 1.00, recall 1.00, against FEMA USA Structures (ORNL footprints; public ArcGIS view, no key).
Old Farm Building: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor appraisal records (public map services, no key).
Old Landmark Building: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor appraisal records (public map services, no key).
Orphan Well Liability Join: proven October 16, 2026 on 24 test cases (3 in the corridor, 10 in Virginia, 11 elsewhere in the U.S.), precision 1.00, recall 1.00, against USGS Documented Unplugged Orphaned Oil and Gas Well Dataset (public ArcGIS, no key).
Opportunity Zone Redesignation Join: proven October 16, 2026 on 24 test cases (8 in the corridor, 16 in Virginia), precision 1.00, recall 1.00, against HUD Opportunity Zones (Treasury-certified 2018 tracts; public ArcGIS, no key).
Pore Space Sequestration Join: proven October 16, 2026 on 24 test cases (14 in the corridor, 1 elsewhere in Colorado, 3 in Virginia, 6 elsewhere in the U.S.), precision 1.00, recall 1.00, against USGS National Assessment of Geologic CO2 Storage Resources (energy.usgs.gov map services, no key).
Private Airfield Distress Join: proven October 16, 2026 on 24 test cases (9 in the corridor, 15 in Virginia), precision 1.00, recall 1.00, against FAA airport master records (5010) via BTS NTAD Aviation Facilities (public ArcGIS, no key).
Quarry Permitted Reserve Join: proven October 16, 2026 on 24 test cases (8 in the corridor, 16 in Virginia), precision 1.00, recall 1.00, against USGS Mineral Resources Data System (WFS mrds-high, no key).
Rail Abandonment Reversion Join: proven October 16, 2026 on 24 test cases (12 elsewhere in Colorado, 12 in Virginia), precision 1.00, recall 1.00, against USDOT BTS North American Rail Network lines (NTAD; public ArcGIS, no key).
Rail Spur Site Join: proven October 16, 2026 on 24 test cases (7 in the corridor, 17 in Virginia), precision 1.00, recall 1.00, against USDOT BTS North American Rail Network lines (NTAD; public ArcGIS, no key).
Rooftop Site Join: proven October 16, 2026 on 24 test cases (1 in the corridor, 23 in Virginia), precision 1.00, recall 1.00, against FEMA USA Structures (ORNL footprints; public ArcGIS view, no key).
Rooftop Solar Site Join: proven October 16, 2026 on 24 test cases (4 in the corridor, 20 in Virginia), precision 1.00, recall 1.00, against FEMA USA Structures (ORNL footprints; public ArcGIS view, no key).
Salvage Yard Distress Join: proven October 16, 2026 on 24 test cases (5 in the corridor, 19 in Virginia), precision 1.00, recall 1.00, against EPA ECHO facility search (public REST, no key).
School Closure Conversion Join: proven October 16, 2026 on 24 test cases (6 in the corridor, 18 in Virginia), precision 1.00, recall 1.00, against NCES public school locations, 2022-23 and current (public ArcGIS, no key).
Seasonal Water (Wetland) Join: proven October 16, 2026 on 24 test cases (17 in the corridor, 7 in Virginia), precision 1.00, recall 1.00, against USFWS National Wetlands Inventory (public ArcGIS, no key).
Ski Area Distress Join: proven October 16, 2026 on 24 test cases (1 in the corridor, 11 elsewhere in Colorado, 12 in Virginia), precision 1.00, recall 1.00, against OpenStreetMap (landuse.
Solar Site Screen Join: proven October 16, 2026 on 24 test cases (6 in the corridor, 11 elsewhere in Colorado, 7 in Virginia), precision 1.00, recall 1.00, against USGS NLCD 2021 (MRLC map service), USGS 3DEP (slope function) and HIFLD substations (public, no key).
Stalled Permit Join: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against City of Boulder construction permits and the Boulder County parcel layer (public ArcGIS, no key).
Stormwater Fee X Impervious Join: proven October 18, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against NLCD 2021 percent developed imperviousness (MRLC, public domain, no key).
Substation Shadow Power Join: proven October 16, 2026 on 24 test cases (5 in the corridor, 19 in Virginia), precision 1.00, recall 1.00, against HIFLD electric substations (public copy on ArcGIS Online, no key).
Truck Parking Desert Join: proven October 16, 2026 on 24 test cases (5 in the corridor, 4 elsewhere in Colorado, 8 in Virginia, 7 elsewhere in the U.S.), precision 1.00, recall 1.00, against FHWA National Highway System Interstates and the Jason's Law truck stop parking file via NTAD (public ArcGIS, no key).
UST Fund Eligibility Join: proven October 16, 2026 on 24 test cases (9 in the corridor, 15 in Virginia), precision 1.00, recall 1.00, against EPA UST Finder, leaking underground storage tank releases (public ArcGIS, no key).
Virginia Jurisdiction Check: proven October 18, 2026 on 9 test cases (1 in the corridor, 7 in Virginia, 1 elsewhere in the U.S.), precision 1.00, recall 1.00, against Census TIGER/Line 2024 Virginia boundary stored in the repository; points at well-known places.
Water System Distress Join: proven October 16, 2026 on 24 test cases (2 in the corridor, 22 in Virginia), precision 1.00, recall 1.00, against EPA Community Water System Service Area Boundaries (public ArcGIS, no key) and EPA SDWIS violations (Envirofacts, no key).
Well Drill Feasibility Join: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Colorado Division of Water Resources well permits and completion reports (CDSS REST API, no key).
Wildfire Insurability Join: proven October 16, 2026 on 24 test cases (18 in the corridor, 6 in Virginia), precision 1.00, recall 1.00, against USDA Forest Service Wildfire Hazard Potential 2023, classified (federal GeoPlatform image service, no key).
Wind Lease Siting Join: proven October 16, 2026 on 24 test cases (5 in the corridor, 7 elsewhere in Colorado, 12 in Virginia), precision 1.00, recall 1.00, against U.S. Wind Turbine Database, USGS/LBNL/ACP (public API, no key).
Zombie Subdivision Plat Join: proven October 16, 2026 on 24 test cases (24 in the corridor), precision 1.00, recall 1.00, against Boulder County Assessor appraisal records (plat, values and parcel tables; public map services, no key).