For lenders, insurers & institutional investors

The NYC distress score, validated out-of-sample to predict what happens next.

OpenStoop grades every NYC multifamily building A–F from 26 official data sources, links each one to other buildings sharing the same registered owner name, and monitors for change every week. Then we did what few in this space do: we backtested the score out-of-sample to show it predicts distress before it happens.

Collateral Watch

Your collateral list, monitored nightly.

Send us your building list (BBLs or addresses). You get a scored watch-list — grades, open Class C hazards, vacate orders, lien-sale events, litigation — refreshed nightly, with alerts when a building deteriorates. NY DFS guidance directs lenders to review outstanding housing-code violations on rent-stabilized collateral; this is that review, continuous.

Validated out-of-sample: in a Feb 2026 held-out backtest (scoring model v5), an F-graded building was ~45× more likely than an A to develop a serious hazardous violation over the following ~10 weeks (AUC 0.83). The current model is re-validated on the same point-in-time protocol weekly. As far as we know, the only NYC building score that publishes a validation like this.

0.83
AUC, held-out backtest (model v5, Feb 2026)
~45×
F vs A odds of a serious hazardous violation over the following ~10 weeks (same backtest)
858k
buildings tracked
445k
graded A–F
26
official NYC data sources
140M+
records, refreshed daily
The wedge

The risk you can’t see from the outside.

Aerial imagery and AVMs tell you what a building looks like. They can’t see the violation, complaint, and litigation history inside it — the record that flags future maintenance-driven distress. OpenStoop reads all 26 official NYC sources, grades every multifamily building against its true peers, ties it to the owner behind it, and shows you which buildings are moving the wrong way — before the habitability problem, the inspection, or the listing.

Lenders & servicers

Early warning on deteriorating building condition — in the public record early, before it shows up in payment performance.

You usually learn a property is in trouble when the payment is late. By then the collateral has been sliding for quarters — in the public record the whole time.

OpenStoop flags deteriorating collateral condition early, and surfaces owner-portfolio contagion the moment a distressed sponsor's other buildings start to slide.

Insurers & MGAs

The habitability risk imagery can't see.

Aerial imagery sees the roof. It can't see the habitability risk inside the building — the violation and complaint history that actually drives habitational loss.

OpenStoop gives you each building's full code, complaint, and litigation history, scored and risk-ranked — so you can prioritize which buildings to inspect and monitor.

Investors & operators

Off-market distress, before it lists.

By the time a distressed building lists, you're bidding against everyone who reads the same listing feed you do.

Source distressed buildings — and the owners behind them — off-market, ranked by distress, and linked to other buildings under the same registered owner name.

How it works

One score. The whole public record behind it.

01

Ingest 26 official NYC sources

26 NYC data sources — DOB, HPD, ECB, FDNY, ACRIS, tax liens, housing litigation, facade (FISP), and more. 140M+ records, refreshed daily.

02

Grade against true peers

Every multifamily building scored 0–100 and graded A–F against buildings of similar size and age in its neighborhood — not the whole city.

03

Link the whole portfolio

Each building is linked to others sharing the same registered owner name, so a distressed sponsor's linked buildings surface together.

04

Monitor for change weekly

New violations, complaints, and filings are tracked weekly, so a building sliding toward distress surfaces before it becomes your problem.

How we validated it

Built to survive your risk team’s questions.

A distress signal is only worth acting on if it predicts out-of-sample. So we tested ours the way you’d test any model before you rely on it.

Out-of-sample, not fit-to-noise

We scored buildings as of a point in time, then measured what actually happened over the following ~10 weeks on held-out data. It was never shown the outcomes it's judged on.

The result: ~45× separation

In the Feb 2026 backtest (model v5), an F-graded building was roughly 45× more likely than an A to develop a serious hazardous violation over the following ~10 weeks. AUC 0.83 — strong, real-world discrimination. The current model is re-validated weekly on the same protocol.

Monotonic — the ranking behaves

Risk rises smoothly from A to F with no reversals. The grade means the same thing everywhere on the scale.

Discriminates even within clean buildings

It separates risk among buildings that have no open violations today — catching the ones on the way down before the first flag lands.

Read it as a ranking, not a verdict. The grade tells you which buildings to watch first. Because of base rates, incidents still come from across the range — a good score isn’t an all-clear, it’s a lower place in the queue. What the score does better than anything else available is sort the haystack.

Methodology note: these figures are from a point-in-time backtest of the score version live at test time. We’re revalidating against the current production model; we’ll share updated figures when it completes.

Pricing

Priced for the desk that needs to act on it.

The score, the portfolio monitoring, and the raw data feed — pick the one that fits your workflow. Start today; cancel anytime.

Analyst
Deal-sourcing & spot checks
$99/mo
  • Search 445k buildings by grade, distress signal, owner
  • Watch up to 100 buildings — daily distress alerts
  • CSV export (metered)
  • Owner-portfolio lookup
Get access
Most popular
Portfolio
Underwriting & monitoring
$499/mo
  • Everything in Analyst
  • Upload your book — batch-score + monitor every building
  • REST API access
  • Owner-portfolio graph + change alerts
  • Reason codes on every grade
Get access
Data feed
Insurers, lenders, platforms
Custom
  • Full bulk data feed (CSV / Parquet) or high-volume API
  • The validated distress score for pricing & reserving
  • Portfolio match-and-append
  • SLA + support + methodology docs
Talk to us

Early access — we’re onboarding the first cohort of lenders, insurers, and investors now. Tell us your use case below and we’ll set you up.

Get a free portfolio audit

Send us a list of addresses or an owner name. We’ll return a graded, risk-ranked report — every building A–F, the distress signals behind each grade, and the owner’s wider portfolio. No account, no billing, no obligation.

Prefer to start small? Look up a single building free.

Prefer to read first? See the methodology or the API docs.

OpenStoop is editorial opinion built from public records — not a consumer report, and not to be used for tenant screening, credit, insurance eligibility, employment, or any other decision about an individual covered by the Fair Credit Reporting Act. It is a property-level risk signal for commercial due diligence. OpenStoop is not affiliated with or endorsed by the City of New York.