{
"key": "<string>",
"n": 0,
"volume": 0,
"n_purchase_money": 0,
"n_investor": 0,
"rank": 0,
"share": 123,
"partial": true,
"floor": true,
"as_of": "<string>",
"dated": true
}The Lender ranking bucket
One place or month a lender lent in: its loans, rank and share there among every lender of the market, and a partial-window flag.
GET /v1/lenders/{id}/rankings: one place or one calendar
month, and what one lender did there over the window the request named. A place is a ZIP, a city, a county or an H3
cell. The ranking row is one lender inside one place. The bucket is one place
inside one lender.
Read a bucket
keynames the bucket: the ZIP, the city, the county FIPS, the H3 index as hex, or the month asYYYY-MM.nis every instrument of any kind the lender recorded in the bucket over the window. The API counts each instrument once across the open-lien and recorded-history tables.volumesums the priced ones. It counts a blanket loan once and leaves out amounts above $5,000,000. That is the same rule as every volume on the Lender object.rankis the lender’s position among every lender of the market in the same bucket and window, byn, with ties by volume then id. Persons below the person gate and placeholder names hold a slot.shareisnover every lender’s instruments in the bucket. On a month bucket, the pair equals the profile’sby_month[]rankandsharefor that month, computed the same way. On a month the lender recorded nothing in,rankis null andshareis 0. The month series fills in such a month so that the series is complete.n_purchase_moneyis the count of instruments amongnthat financed a purchase.n_investoris the count of those whose borrower resolves to a registered investor. Both are null where no deed link exists for the market.partialis true when the bucket is the slice month or the slice year, a window that ends afterrecordings_through. The bucket is the slice month when its key is that month, or whenperiod=YYYY-MMnames that month. The bucket is the slice year whenperiod=YYYYnames that year.flooris true on a calendar-year bucket beforehistory_floor_year. The history keeps five slots per parcel, so those years are a floor, never a count.
as_of is the registry’s slice date, and meta.dated[] on the response names the lenders
and financing blocks. A bucket carries no name and no address. The lender is the path id.One place, month or cell the lender lent in over the period. The bucket carries its count, volume, purchase-money and investor counts, and its rank and share among every lender there.
The bucket, by group_by. A ZIP is 5 digits. A city is upper case as the parcel table writes it. A county is its 5-digit FIPS. A cell is an H3 hex string, for example 8848eba59dfffff. A month is a calendar month, YYYY-MM.
Instruments recorded in the bucket over the period with this lender as lender of record. The count takes each instrument once across the open-lien and recorded-history tables. 0 on a month of the span in which the lender recorded nothing.
-9007199254740991 <= x <= 9007199254740991Sum of the loan amounts of the priced instruments in the bucket over the period, whole dollars. The sum counts a blanket group once and leaves out amounts above $5,000,000. 0 when none carries a price.
-9007199254740991 <= x <= 9007199254740991Of n, the instruments whose purpose is PURCHASE. A PURCHASE has a priced deed on the parcel in the 45 days up to the recording. Null while the market has no deed link (purchase_measured false).
-9007199254740991 <= x <= 9007199254740991Of n, the instruments whose borrower resolves to a registered investor. Null while the market has no borrower match.
-9007199254740991 <= x <= 9007199254740991The lender's rank by n in the bucket over the period among every lender active there. The ranking takes every class, persons below the gate and placeholders included. It breaks ties by volume, then by id. Null on a month of the span with n 0.
-9007199254740991 <= x <= 9007199254740991n over every lender's instruments in the bucket over the period, 0 to 1. The denominator counts every lender. 0 on a month with n 0.
True when the bucket's window ends after recordings_through. That is the slice month, as a month bucket or as period YYYY-MM, and the slice year, as period YYYY. Recordings in the slice stop on recordings_through.
True on a calendar-year period before history_floor_year. The recorded history keeps five slots per parcel and pushes older mortgages out, so the count is a floor.
The slice date, YYYY-MM-DD. The API measures the counts at it. Dated: see meta.dated[].
True while the datasets are not in the current delivery, so the row is a snapshot at as_of.
Was this page helpful?