The intelligence graph connecting private capital, companies and real assets. Public records ingested on their own cadence, resolved to one identity by shared identifier, connected by typed relationships with a date and a source on every edge, and read by detectors that say what changed and what it means. Facts, derived metrics, signals, anomalies and inferences are labelled as such.
Reading the graph
Pick states, cities or counties, property types, a loan size and the lenders you care about. Every week DFX compares the public FHA and CMBS loan tape with what it saw the week before and sends the few changes that matter: loans entering the 12, 18 or 24 month window, maturity dates that moved, lenders of record that changed, loans that left the tape, HUD contracts and LIHTC compliance periods coming due. In New York City it also reads the recorded mortgages of banks, life companies and debt funds: new mortgages, refinancings, assignments and payoffs. Each item names the property, the loan, who originated or holds it, the owner or borrower where the record is reliable, and who to call.
The largest loans that entered a maturity window in the last seven days, from every state. A saved universe narrows this to your market and adds the changes below the headline.
Loans that are never securitised or insured never reach a tape, but in New York City every one is recorded with the City Register. DFX reads each monthly ACRIS edition for multifamily (five units and up), office, retail, industrial, hotel and mixed use properties with $1M or more secured, and names the lender, the borrower of record and the parcel. A recording carries no maturity and no balance: amounts are what was secured when the document was recorded, and the aged mortgage list is an inference from loan age, never a maturity date.
| Recorded in | Commercial mortgage recorded | Loan consolidated, modified or extended | Mortgage assigned to a new holder | Mortgage satisfied (paid off) |
|---|---|---|---|---|
| 2026-08 | 226 ($2.8B) | 355 ($5.1B) | 478 ($10B) | 105 ($1.3B) |
| 2026-07 | 327 ($3.5B) | 370 ($5.6B) | 513 ($12B) | 94 ($1.6B) |
| 2026-06 | 296 ($3.3B) | 430 ($9.0B) | 570 ($12B) | 76 ($456M) |
| 2026-05 | 201 ($2.3B) | 262 ($6.1B) | 457 ($8.6B) | 73 ($775M) |
| 2026-04 | 216 ($2.7B) | 390 ($8.4B) | 550 ($15B) | 65 ($562M) |
| 2026-03 | 282 ($2.5B) | 337 ($7.1B) | 497 ($8.7B) | 95 ($1.0B) |
10,254 commercial mortgages recorded five or more years ago have nothing later on the record (no satisfaction, assignment, modification, new mortgage or deed on their lots). Source: NYC Department of Finance, ACRIS real property master, parties, legals and references on NYC Open Data, published monthly; building class from NYC PLUTO. Choose the New York City shortcut in the form below to include these in a universe.
A universe drawn around a few buildings is empty most weeks. Start from every multifamily and office loan over $5M in your states, a lender's whole book, or the HUD contracts expiring in your states, then narrow it. Each preview shows the last four weeks and how many items a week the universe would have carried over the last 13 weeks. The preview is free and needs no account. Saving needs a free DFX account and starts a weekly digest on Monday; the first one covers the last four weeks. A week with nothing material sends nothing.
Each change family is checked against a blind labelled sample. Only families validated for alerts reach a paid digest; public signals appear labelled in free digests, and withheld families are never sent.
| Family | Status | Precision | Sample |
|---|---|---|---|
| Aged commercial mortgage, no satisfaction (NYC, inference) | Validated for alerts | 100% | 30 |
| Commercial loan modified, consolidated or extended (NYC) | Validated for alerts | 100% | 30 |
| Commercial mortgage assigned (NYC) | Validated for alerts | 100% | 30 |
| Commercial mortgage recorded (NYC) | Validated for alerts | 100% | 30 |
| Commercial mortgage satisfied (NYC) | Validated for alerts | 100% | 30 |
| Distress status change | Public signal | 80% | 30 |
| Enforcement signal | Withheld | 48% | 25 |
| Lender or servicer change | Not yet measured | n/a | n/a |
| LIHTC compliance period end | Public signal | 100% | 30 |
| Maturity date changed | Public signal | 80% | 15 |
| Loan enters 12-month maturity window | Validated for alerts | 100% | 30 |
| Loan enters 18-month maturity window | Validated for alerts | 97% | 30 |
| Ownership change on a financed property | Not yet measured | n/a | n/a |
| Subsidy contract expiration | Validated for alerts | 100% | 30 |
How a change is known is printed on every item: seen changing between two DFX snapshots of the tape, computed from a filed maturity or expiry date, or a published servicer event. CMBS lender names are the originators on the filing; the note itself sits in a securitisation trust.
The loan tape names the property, the lender and the maturity, never the borrower. DFX finds the owner of record by an identifier join or an identical address with a uniqueness condition, never by matching names, and each way of finding it is checked by hand against the public record. A basis names the owner only where at least 50 loans were checked and 95% of them were right; a weaker basis is labelled unverified, and one weaker still is not printed at all. A natural person is never named.
| Where the owner comes from | On an item | Right | Loans covered |
|---|---|---|---|
| HUD multifamily owner record for this FHA loan | Named as the owner | 100% of 60 | 4,200 |
| mortgagor on the recorded mortgage for this loan | Named as the owner | 100% of 60 | 288 |
| HUD multifamily owner record for this property | Named as the owner | 100% of 60 | 40 |
| New York City tax roll owner for this lot | Named as the owner | 98% of 60 | 269 |
| assessor record at the same address | Named as owner of record (unverified) | 82% of 60 | 36 |
| grantee on the latest NYC deed DFX holds for this tax lot | Not printed | 66% of 32 | 55 |
An owner of record on a financed building is nearly always a single purpose company, so DFX also looks for the party above it, and only in a filing that states the relationship: HUD's owner participant record, or the head officer a New York City building owner has to register every year. A registered agent, a law firm and a managing agent are refused, because none of them controls the borrower. The same bar applies: at least 50 checked by hand and 95% right to say controlled by, at least 30 and 80% to name the filing, nothing printed below that.
| Where the controller comes from | On an item | Right | Properties covered |
|---|---|---|---|
| named on the building's New York City HPD registration | Named on the filing (unverified) | 93% of 60 | 155 |