Search properties with filters

Owner's mailing address (street + 5-digit zip, case/punctuation-insensitive) differs from the property's site address — an investor / non-owner-occupied signal. true = absentee, false = owner mails to the property; parcels with no mailing or site address match neither. Evaluated per candidate document (no stored flag yet), so always pair it with a geography filter (state, county, zipCode, fips, a radius) — an unscoped absenteeOwner over all parcels may time out.

Mortgages believed to still be outstanding (e.g. {lte: 0} for parcels with none). {lte: 0} alone is NOT a free-and-clear filter: it also matches recently-sold parcels whose new mortgage has not been recorded yet. Pair it with preSaleLoanCount: {lte: 0} for parcels we can actually verify.

Parcel sits in a qualified census tract for affordable-housing programs (17,603,810 parcels). Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Agent id(s) (GET /v1/agents/{id}): parcels with a recorded sale by the agent.

Confidence in the value estimate, as a 50–100 score (e.g. {gte: 90} for high-confidence estimates only). NOT a fraction — {gte: 0.9} matches every scored parcel, not the top decile.

Automated value estimate for the parcel, in whole dollars (e.g. {gte: 500000}). ~77% coverage. Matches the avm.mid value on the response.

Bedroom count (e.g. {lte: 2} for starter homes). ~46% coverage — on the rest the bedroom count was never recorded, which is NOT the same as a parcel with no bedrooms and NOT the same as a parcel with no assessor record: 15.6M of them do carry a recorded bath count and floor area. Unrecorded parcels are EXCLUDED by default; pass includeNulls: true to add them back. beds: 0 matches nothing (an unrecorded count is not a count of zero, and a genuine studio is indistinguishable from it), and the response returns null rather than 0.

County, as a 5-digit FIPS code (e.g. 06037). Canonical key for the county dimension; fips is the exact-code alias kept for back-compat.

Parcel sits in a HUD-designated difficult development area (30,643,419 parcels). Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Parcel sits in a distressed-or-underserved nonmetropolitan middle-income tract (13,289,914 parcels). Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Estimated equity, in whole dollars (e.g. {gte: 250000}). Negative values are genuine underwater parcels. Derived from the value estimate minus outstanding loan balances, so it inherits their coverage (~77%) AND their gaps: on a parcel whose mortgage is missing from the record there is nothing to subtract, so this returns the full value estimate. Pair a high-equity range with preSaleLoanCount: {lte: 0} to drop those.

Estimated equity as a FRACTION of the value estimate, where 1 = the whole value and 0.5 = half. Pass {gte: 0.5} for 50%-or-more equity — {gte: 50} is not a synonym and matches ZERO parcels, since no value exceeds 1. Negative values are genuine underwater parcels. Note the contrast with loanLtv, which takes whole percents. CAUTION: 1 is not proof a parcel is owned free and clear — it is also what a recently-sold parcel returns when the buyer's new mortgage has not been recorded yet, because the calculation subtracts nothing. Add preSaleLoanCount: {lte: 0} to restrict to parcels whose position we can actually verify.

Parcel ids to exclude — the inverse of id.

Parcel sits in a FEMA special flood hazard area (4,675,160 parcels). tractFloodZone gives the specific zone code. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Recording date of the most recent foreclosure notice on the parcel (e.g. {gte: "2025-01-01"}). Parcels with no notice are excluded. Pair with foreclosureStage, e.g. {foreclosureStage: "ND", foreclosureDate: {gte: "2025-01-01"}}.

Stage code of the most recent recorded foreclosure notice on the parcel (e.g. "ND", "NS"). Present on every parcel whose foreclosureStatus is in_foreclosure, so any value here also limits results to those parcels. Accepts one code or an array (any-of). Pair with foreclosureDate to find recent notices of one stage.

Map viewport {north, south, east, west} in degrees: only listings inside the box. west greater than east crosses the antimeridian. Unlike geoPoint (a radius) this is the shape a map reports. Applies to list, count and analytics alike.

Radius search — { lat, lon, radius } (e.g. radius "15mi" or "24km") — around any office on the agent's record; with locationSource office the CURRENT office, production their transactions.

Exact parcel ids (up to 50,000, any-of). A property id from the list/search rows (mm_<fips><apn>|<fips>-<apn>|<unit>, exactly that parcel) or an mmPropertyId (mm_<fips><apn>, every unit of the parcel). Must start with mm_. ANDed with the other filters, even under mode: "or".

Per-loan funding-company NMLS id (digits only) on the parcel's loans.

Funding-company name on the loan, case-insensitive contains.

NMLS id matched as the loan's funding company OR lender.

Loan filters to EXCLUDE — each key negated on its own, under the same loanMatchMode. With "most_recent", {loanCompanyOrLender: "3030"} keeps parcels whose newest loan is not from 3030 (no-loan parcels included); {loanRecordedDate: {lt: "2017-01-01"}} drops newest loans recorded before 2017, keeping parcels with no date.

Deprecated alias of loanModeledInterestRate — a low-biased ModelMatch model that caps near 7%. For the genuine recorded note rate use recordedInterestRate.

Lender ID on the parcel's loans — the normalized lender-dictionary key, NOT the lender name the response shows. Resolve a lender via instantSearch (or listLenders) and pass the id it returns (e.g. "mortgage_rocket" or "mortgage rocket"); the lenderId on a parcel's loans[] is the same value, so a lender read off a result round-trips here directly. Underscores and letter case are normalized, and the id expands to every stored variant the lender is recorded under. A value that is neither an id nor a stored lender name returns 400 with the closest matches, rather than silently matching nothing — to filter by the lender NAME shown on the parcel, use loanLenderName.

Lender NAME on the parcel's loans — the counterpart to the exact-id loanLender, and the filter that accepts the lenderName a parcel's loans[] displays (e.g. "UNION HOME MORTGAGE CORP"). Case-insensitive, matched against the name recorded on the loan. When the name is one the lender dictionary knows, it also expands to that lender's whole recorded-name family, so a brand name ("union home mortgage") selects far more than the one literal string. Never 400s: a name it cannot resolve simply matches the recorded string. The response reports how each value resolved as lenderNormalizations.

Lender name on the loan, case-insensitive contains (loanLenderName is the exact / name-family match).

Loan-to-value on the parcel's active loan(s), as WHOLE PERCENTS — {gte: 40, lte: 50} means 40%–50% LTV. A fractional range ({gte: 0.4, lte: 0.5}) is not a synonym and matches ZERO parcels. ~7% coverage. Careful: the response DTO reports the same measurement as a fraction (loans[].ltv of 0.4238 is 42.38% LTV), so a value read off a result must be multiplied by 100 before it can be used here. equityPct, by contrast, takes the fraction. Two further limits of the stored value. It SATURATES at 10 and 199 — no parcel stores a ratio outside that range, so {gte: 200} matches nothing at all, and the extremes are dominated by parcels whose true ratio was merely clipped to the bound; the response returns those as null, but this filter still selects them. And the underlying value is not a dependable per-loan ratio — two thirds of parcels with several mortgages store an identical figure across loans of very different amounts (see loans[].ltv, deprecated). For selecting on borrower position prefer equityPct, which has neither limitation.

How the loan-slot filters combine across a parcel's 5 cached loans. correlated (default): ONE loan (any slot) must satisfy ALL loan-slot conditions. most_recent: only the newest loan (loan_1). any: each condition may be satisfied by a different loan. none: ignore loan-slot filters. Honored on list, bulk-delivery, and bulk-enrich.

Per-loan MODELED rate (loan_N.modeled_interest_rate). Low-biased, effectively caps near 7% — not a note rate. Use recordedInterestRate for the recorded rate.

Per-loan originator NMLS id (digits only) on the parcel's loans. For all-time attribution across the parcel use loNmls.

DEPRECATED — fuzzy name matching is unreliable (same originator can return wildly different counts by name string). Resolve the loan officer via instantSearch and filter by loanOriginator (exact NMLS id) instead. Still functional for now.

Loan officer name on the loan, case-insensitive contains. Prefer loanOriginator (NMLS id).

Lot size in square feet (e.g. {gte: 5000}). ~85% coverage; unmeasured parcels are excluded by default — pass includeNulls: true to add them back. lotSquareFeet: 0 matches nothing (the response returns null).

Parcel sits in a low- or moderate-income tract under the Community Reinvestment Act (34,010,153 parcels). tractIncomeLevel is the four-way form of the same classification. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Parcel sits in a majority-minority tract (38,745,783 parcels). Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Straight-line distance from the parcel to the nearest college, in miles. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Straight-line distance from the parcel to the nearest school, in miles. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Straight-line distance from the parcel to the nearest place of worship, in miles. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Parcel sits in a federally designated Opportunity Zone tract (10,277,168 parcels). Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Assessor owner-occupancy flag, bucketed: "owner_occupied", "probable_owner_occupied", or "not_flagged" (the assessor did not flag the parcel — where non-owner-occupied/investor parcels land). Accepts one value or an array (any-of). For investor targeting use {ownerOccupancy: "not_flagged", absenteeOwner: true}, optionally with ownerType: "entity".

Owner vesting type (any-of); individual = neither entity nor trust.

Mortgages excluded from activeLoanCount because the parcel sold after they were recorded — the seller's loan, assumed paid off at closing. Use it to separate the two meanings of activeLoanCount: 0: {activeLoanCount: {lte: 0}, preSaleLoanCount: {lte: 0}} returns parcels where nothing was excluded, so the zero is backed by the record; {activeLoanCount: {lte: 0}, preSaleLoanCount: {gte: 1}} returns the parcels where it is not. On its own a non-zero value is unremarkable — most recently-sold parcels have one and also carry a current mortgage.

Restrict to parcels whose value estimate falls within this loan program's county price limit — an eligibility band (value ceiling). Pair with fips to resolve the county's limit; with no county context the national baseline applies. Parcels without a value estimate are excluded. This is a price band only — no borrower income, payment, or DTI is considered.

Genuine RECORDED note rate from the parcel's active loan(s) (~7% coverage, concentrated on the most-recent loan). Percent input (e.g. 7 means 7%).

Parcel sits in a rural tract (29,289,319 parcels). Thinner coverage than the other flags — stamped on 111.9M of 161.1M parcels, and a parcel with no value is dropped by either true or false. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Likelihood the parcel sells in the near term, as a 0–100 score (e.g. {gte: 80} for the most likely to sell). NOT a fraction — {gte: 0.8} matches nearly every scored parcel. salePropensityCategory is the bucketed form of the same signal.

Building area in square feet (e.g. {gte: 1500}). ~68% coverage; unmeasured parcels are excluded by default — pass includeNulls: true to add them back. squareFeet: 0 matches nothing (the response returns null).

Mortgage applications reported in the tract over the year, as a count. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Asian (non-Hispanic) share of the tract's population, 0-100. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Black (non-Hispanic) share of the tract's population, 0-100. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Three-digit county FIPS code of the tract (county-within-state, NOT the 5-digit county/fips key). Pass with tractStateCode and tractId to pin one neighborhood. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Share of the tract's mortgage applications that were denied, 0-100 (e.g. {gte: 25} for 25%+). Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Median family income of the tract, in whole dollars (e.g. {lte: 60000}). Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Hispanic or Latino share of the tract's population, 0-100. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Owner-occupied share of the tract's housing units, 0-100. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Median household income of the tract, in whole dollars (e.g. {gte: 120000}). Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Six-digit tract code. A tract code REPEATS in every state and county, so this only identifies a neighborhood when passed ALONGSIDE tractStateCode and tractCountyCode — on its own it matches roughly 75× too many parcels. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Tract income relative to its metro area, where 100 is parity — {lt: 80} is the CRA low/moderate band and values legitimately exceed 100 (observed up to 412). NOT a 0-100 share. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Mortgage dollar volume reported in the tract over the year, in whole dollars. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Median age of the tract's population, in years. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Median home value across the tract, in whole dollars. A neighborhood statistic — the parcel's own value estimate is avmValue. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Non-white share of the tract's population, 0-100 (e.g. {gte: 50} for majority-minority; majorityMinorityTract is the flag form). Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Mortgage originations reported in the tract over the year, as a count. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Total population of the tract. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Share of the tract's population below the poverty line, 0-100 (e.g. {gte: 20} for 20%+). Pass whole percents — {gte: 0.2} means 0.2% and matches almost every tract. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Remoteness of the tract on an ORDINAL 1-10 scale (1 = metropolitan core, 10 = most remote). This is a code, not a percent — {gte: 7} selects the most remote tracts. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Two-digit state FIPS code of the tract. Pass with tractCountyCode and tractId to pin one neighborhood. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

White (non-Hispanic) share of the tract's population, 0-100. Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

Parcel sits in a tract eligible for USDA rural housing programs (70,637,816 parcels). Values describe the U.S. Census tract (neighborhood) as a whole — tract-level aggregate statistics, never an individual, household or applicant attribute. Percentages are 0-100.

ZIP code. Canonical key (matches loans zipCode and the response DTO); zip is a legacy alias kept for back-compat.

Opaque cursor from previous response for next page

Page size (1-100, default 25)

Where rows without a value go (default last in both directions). first with asc / last with desc treats blanks as the smallest value.