Skip to main content
Early access: while the developer tier is in beta, the API serves the routes on this page to Investorlift’s team and trusted partners. Investorlift will restrict them further before they open to every key.
Every recorded mortgage names a lender. Since 0.14.0 every lender of record carries a stable id (len_ plus 12 hex). The id is the same under every spelling the instruments use, and in every market. Kiavi is the worked case. The county recorded its loans under five spellings: KIAVI FUNDING INC, KIAVI FUNDING LLC, LENDINGHOME FUNDING CORP, LENDINGHOME FUNDING CORPORATION and LENDING HOME FUNDING CORPORATION. The registry folds the five into one lender. Read two limits before any number. The registry is a snapshot as of meta.coverage[].lenders.as_of, with recordings through meta.coverage[].lenders.recordings_through. meta.dated[] says so on every response, and dated data explains the stamp. The registry covers the counties in meta.coverage[].lenders.counties. Counties, cities and ZIP codes lists them. The recorded history captures about one open loan in five (coverage.history_capture_share). So every count here is over the union of the open liens and the history, and a history count is a floor. A bridge lender is the most undercounted lender of all, because its loans are short and the borrower pays them off before the next slice. Compare Kiavi with the other hard-money lenders, not with a bank.

​
Find the lender

The hit carries name KIAVI FUNDING INC, matched_name LENDINGHOME FUNDING CORP and match: tokens. The spelling you searched is one of the lender’s, and the id is the lender’s. is_hard_money is true by name (hard_money_basis: NAME), and lender_class reads NONBANK. A search for “Kiavi Funding, Inc.” matches exact. Case, punctuation and the entity suffix never matter.

​
Read the profile

names[] lists the five spellings, the display name first. former_names[] lists the three the dictionary retired under the rename. markets[] carries one profile, Phoenix. In it:
  • n_loans_24m is the origination count over the 24 months to as_of, counted once across both tables. n_in_history_24m is what the history holds, and it is smaller. n_open_liens and open_balance are the open book at the slice date. The balance is after the exclusions that the Lender object lists.
  • rankings.rank_24m and share_24m place the lender among every lender of the market. rank_hard_money_24m and share_of_hard_money_24m place it among the hard-money lenders. That comparison is the valid one.
  • terms says what it lends: the amount quartiles over the window, and the rate and term figures. Beside the rate and term figures it says how much of the book carries a rate or a term. On a bridge lender’s book a term is on about one row in ten (term.share_term_known). So the term bands and product_proxy describe a minority, and the API serves them with that share beside them.
  • geography.by_zip and concentration_top3_zips say where. by_month gives the lender’s count against the market’s each month, with its rank. The last month is partial.
  • by_purpose_24m splits the window by what the deeds say each loan did (PURCHASE, NOT_PURCHASE, UNKNOWN). rankings.rank_purchase_24m places the lender by its purchase-money loans. terms.loan_to_price is the loan against the deed price on those loans. A median above 1 is a rehab holdback on top of the price. The three blocks below say who borrows and what the loans financed.
  • Since 0.19.0 rankings also counts the places the lender leads: n_zips_ranked_first_24m and n_cities_ranked_first_24m. Those are the ZIPs and cities where two facts hold. The lender recorded 10 or more loans there in the window, and it ranks first there among every lender of the market. takebacks says what it took back at foreclosure since 2016, by instrument, with is_auction_lender. The take-backs are deeds from the county’s records, never delinquency. The block is null as a whole where the market has no take-back join (takebacks_measured).
  • summary is one paragraph written from those numbers at build time. It never quotes a margin.

​
List the loans

The list puts the newest recording first, one row per instrument. The row is the open row where the instrument has one, else its recorded slot (source, in_both). Each row carries the parcel, and property_id opens its financing block. It also carries the amount, balance, rate, term and maturity as filed, and product, term_band and product_proxy read from them. A row with term_band: UNKNOWN carries no proxy. lender_name_as_recorded is the spelling on that instrument. The borrowers of a recorded row are a contact field, and this host does not put them on the row. The same request with Accept: text/csv returns the whole window as lender-len_d2028f0766fc-loans.csv from Growth, up to 50,000 rows. The CSV section has the details. Because of the recorded_from window, the API can answer this request for the busiest lenders too. Since 0.16.0 each row also says what the deeds know about it:
  • purpose is PURCHASE when a priced deed on the parcel is dated in the 45 days up to the recording. The deeds decide it, never the file’s label.
  • deed with loan_to_price is on such a row.
  • deal names the investor purchase linked to the loan. It says if this loan is the purchase loan, and what the deeds show came next.
  • outcome is RESOLVED_BY_RESALE, REFINANCED, OPEN or UNKNOWN. It is a performance proxy read from the deed chain, never a payment history.
  • The borrowers are keys and investor ids.
purpose=PURCHASE keeps the purchase-money loans. deal_kind=flip keeps the ones that funded a flip. investor_id= keeps the loans of one investor. Since 0.19.0 a row also says if a take-back deed followed it: foreclosed, with the instrument and the deed date. outcome reads FORECLOSED on a loan tied to a deal. foreclosed=true keeps those rows. cell= or a point with radius_miles keeps the rows in one hex cell or around a point. The API resolves the geometry to the cells whose centre lies inside it, so its edge is a hex boundary, not the circle.

​
Who borrows, and what the flips made

The profile’s three linkage blocks answer the questions a lender customer asks first. borrowers says how many borrowers the lender has (n_borrowers, n_borrowers_24m). It says how many resolve to a registered investor, how many repeat, how many are new this year, and how many churned (n_churned_12m). A churned borrower’s last loan here is 12 to 36 months old. The borrower took a later loan elsewhere, with none of this lender’s open liens after it. Churn is a lower bound of leaving and an upper bound of nothing. The recorded history sees about one loan in five. So a borrower who stayed can read as gone only when this lender’s later loan is in neither table. churn_measured says if the API can read churn at all. top_borrowers[] lists the ten largest by key. A company’s name is beside its key, and the list never shows a person’s name (person_names_redacted). competing_lenders[] lists the ten lenders that share the most borrowers with this one. investor_lending says how much of the book funds registered investors. It carries share_investor_loans_24m, the investors by kind, and is_investor_lender when half or more of the window goes to them. It also says which deals the lender’s loans financed, by what occurred on them, and names the ten largest investors. by_outcome_24m and median_months_to_payoff say what became of the captured loans. flips_financed is the margin block. It counts the flips bought in the last three years whose purchase loan was this lender’s. capture_share is that count against the flips by investors who borrow from it elsewhere. The block also carries the lender’s share of every flip the slice can attribute. It carries median_gross_profit with its quartiles, median_spread_ratio, median_hold_days and median_loan_to_price where 20 or more of those flips resold with both prices. Under 20 the margins are null. A gross profit is never net of rehab or costs. The capture behind all of it is partial. The open-lien table sees about one purchase loan in five on holdings and one in fifteen on flips. Every block is null where its input does not exist for the market (purchase_measured, investor_lending_measured, flips_measured, borrowers_measured).
GET /v1/lenders/{id}/borrowers lists every borrower, most loans in the window first. Each row carries:
  • borrower_key: a salted hash of the folded name, the same key on every lender’s rows.
  • name, for a company.
  • investor: the registered investor behind the key. Its id opens its profile, whose financing block lists every lender it borrows from.
  • The loans and volume from this lender.
  • other_lender_ids[].
  • The churn triple.
registered_only=true keeps the registered investors. churned=true keeps the churned borrowers. q= finds a company by name. period=all counts the whole history instead of the window. This host does not serve a person borrower’s name. The row carries name null with contact_redacted: true, and the API serves the key and the investor.

​
Who lends to flippers in 85032

The list shows the hard-money lenders active in the ZIP over the window, most loans first. Each row carries n (its loans there), share (of every lender’s loans there), n_first_lien, volume and last_recorded_on. rank is the position in this list. Without hard_money=true the banks and mortgage companies lead the same list. The list hides persons and government lenders unless you ask for them. A ZIP outside the counties the registry covers answers 422 outside_coverage and names the covered counties, never an empty page. Since 0.19.0 period= picks the window: 12m, all, a year such as 2025, or a month such as 2026-05, beside the default 24m. So the same call with period=2026-05 is the ZIP’s ranking for one month, and twelve such calls are its monthly series. A point with radius_miles in place of the ZIP ranks the lenders around it. The row’s in_geometry says how many hex cells the circle became.

​
Which ZIPs does this lender lead

GET /v1/lenders/{id}/rankings (since 0.19.0) is the ZIP list seen from the lender’s side. It returns one bucket per ZIP where the lender recorded 10 or more loans over the window, most loans first. Each bucket carries the lender’s n and volume there. It also carries its rank and share among every lender of the market in that ZIP, persons and government included. The buckets with rank: 1 are the ZIPs the profile’s n_zips_ranked_first_24m counts. The summary says how many ZIP codes the lender ranks first in, from the same number. group_by=city and group_by=county do the same by place. group_by=cell&res=8 does it by hex cell, for the profile’s map, and GET /v1/lenders/{id}/cells gives the counts alone. group_by=month&period=all is the lender’s 36-month series against the whole market, every month present, the last partial. A registry built before the place rankings answers 422 lenders_unavailable here and serves the three ranked-first counts as null.

​
The rules behind the numbers

Lender names are business records of the loan, and the API serves them in full to every key, whatever the class. Nothing in any delivery carries a lender’s phone or email. A company borrower’s name on a recorded instrument is a business record, and the API serves it the same way. This host does not serve a person borrower’s name, and a profile or a list over MCP never carries it. A lender whose name reads INDIVIDUAL or PRIVATE has a profile only above the person gate. The lists and searches hide such a lender unless include_persons=true. No lender route has a price in the beta. The API charges no credits for any of them. A market with no published lender registry answers 422 lenders_unavailable on the seven routes. A registry built before the deed link answers the same on the borrowers route and on the purpose, outcome and investor filters. A registry built before the place rankings answers the same on the rankings and cells routes. It also answers the same on a period other than 24m, on a geometry, and on cell and foreclosed. meta.coverage[].lenders is null where the market has no registry. Where it has one, the block carries the *_measured flags.