METHODOLOGY · SCORING
How buildings are scored.
Every building in the CondosCoopsNYC catalog receives a grade derived from the same formula, the same data sources, and the same thresholds. No building is singled out. No building is given the benefit of the doubt. The formula runs once, uniformly, across every property in the catalog.
This page explains the principles behind the score — what it measures, which public records feed it, and how bias is engineered out. The model that turns those records into a grade is proprietary work product, available to regulators, lenders, auditors, and counsel under confidentiality agreement.
PRINCIPLES
What makes a score fair.
Per-unit normalization
Raw violation counts are meaningless without context. A 500-unit building with 50 violations is better-run than a 20-unit building with 50 violations. Every metric is normalized to a per-100-unit basis before any comparison or threshold is applied.
Same formula, every building
The grading function runs identically on every building in the catalog. There are no manual overrides, no building-specific adjustments, and no editorial discretion in the letter assignment. If you change the data, the grade changes. Nothing else changes the grade.
Public-record inputs only
Every input to the grade is drawn from a government-published dataset: NYC HPD, DOB, OATH/ECB, FDNY, DOF, DHCR, DCP PLUTO, ACRIS, NYSCEF, the Public Advocate, and the NYS Attorney General. No private data. No anonymous tips. No editorial weighting by neighborhood, price point, or ownership structure.
Severity-weighted, not count-weighted
A Class C violation (immediately hazardous: lead paint, no heat, gas leak) carries far more weight than a Class A violation (non-hazardous: peeling paint in a hallway). This reflects the actual risk hierarchy established by HPD's own classification system, not an editorial choice.
Correction-aware
A building that fixes its violations quickly looks different from one that lets them sit open for years. The scoring system tracks median correction time and distinguishes between open and resolved violations. When DOB high-severity violations have been dismissed on appeal, the system treats that as context rather than a penalty. A building that fights a citation and wins should not carry the same weight as one that ignores it.
Verifiable inputs, proprietary model
Every input to a grade is a public record you can pull and check yourself — the violation, the fine, the housing-court case, the FISP status. What is proprietary is the transformation layer that turns those records into a letter: the weights, density bands, and triggers. That model is available to regulators, lenders, auditors, and counsel under confidentiality agreement.
TWO SYSTEMS
Composite score vs. forensic grade.
CondosCoopsNYC uses two complementary scoring systems, each designed for a different purpose.
Composite score catalog-wide
A single numeric score computed for every building in the catalog during data enrichment. Used on the grades page and master building list for at-a-glance comparison and sorting. Higher is worse.
Forensic grade audit reports
A letter grade (A through F with 1-10 sub-tiers) computed at report-generation time using a richer set of inputs including live Socrata queries, OATH hearing outcomes, FISP cycle history, LL97 emissions modeling, and cross-dataset derived signals. Used in the Level 2 and Level 3 PDF audit reports.
COMPOSITE SCORE
The catalog-wide formula.
The composite score is a weighted sum of six public-record inputs. Each input reflects a different axis of building governance failure. The weights are calibrated so that the most dangerous conditions (immediately hazardous violations, housing-court judgments) dominate the score, while lower-severity signals (311 complaints, minor fines) contribute proportionally less.
The exact weights, and the composite-score boundaries that map a number to a letter on the grades page, are proprietary. What is not: the inputs are all public records, and the weighting is strictly severity-ordered — a single immediately-hazardous (Class C) violation or a housing-court judgment moves the score far more than a minor fine or a 311 complaint. A building with a clean public record sits at the bottom of the range; the worst buildings in the catalog score orders of magnitude higher.
FORENSIC GRADE
The audit-report grading system.
The forensic grade used in Level 2 and Level 3 audit reports is a richer, more granular system. It starts from per-unit-normalized density thresholds to assign a base letter, then shifts the letter downward based on specific governance triggers found in the building's record. The result is a letter (A through F) with a sub-tier (1-10) that captures position within the letter.
From density to letter
The forensic grade starts from the building's per-unit-normalized density — Class C and total HPD violations and ECB fines per 100 units — to set a base letter, then shifts that letter based on severe findings in the public record. The exact density bands, and the weight each trigger carries, are proprietary; the triggers themselves are public flags anyone can look up:
- AEP worst-buildings list — HPD's official worst-buildings registry.
- Active vacate order(s) — the city has ordered residents out.
- Public Advocate Worst Landlord Watchlist.
- Repeated immediately-hazardous (Class C) violations.
- HPD housing-court judgments against the landlord.
- Live DOB high-severity violations (not dismissed on appeal).
- Condo board-v.-sponsor or repeated unit-owner litigation.
- Facade UNSAFE under LL11, LL97/LL84 non-compliance, ACRIS liens and similar capital-and-compliance signals.
Grade F is reserved for life-safety crises (such as an active vacate order) or a simultaneous multi-axis collapse. The precise thresholds and the arithmetic that combine these signals into a letter and sub-tier are part of the proprietary model.
THE SCALE
What each letter means.
| Letter | Sub-tiers | Meaning |
|---|---|---|
| A | A-1 … A-10 | Well-governed. Baseline compliance, no major signals. |
| B | B-1 … B-10 | Mostly compliant. Minor issues only. |
| C | C-1 … C-10 | Compliance gaps. Ongoing concerns, mixed signals. |
| D | D-1 … D-10 | Significant governance failures. Multiple operational concerns. |
| E | E-1 … E-10 | Severe distress. Emergency-level compounding failures. |
| F | (none) | Total failure. Life-safety crisis or multi-axis collapse with no recovery path. Avoid. |
RISK DASHBOARD
Per-dimension severity scoring (0-100).
In addition to the composite letter grade, every audit report includes a Risk Dashboard: a set of horizontal severity bars, one per governance dimension, each scored 0-100. The dashboard shows where the risk concentrates. A building might have a clean HPD record but severe LL97 exposure, or vice versa.
Each dimension's score is computed from its own input data, normalized to fit a 0-100 scale using dimension-specific calibration. The normalization ensures that a score of 50 means "meaningfully elevated" and 80+ means "severe" across every dimension, regardless of the underlying unit of measurement.
The dashboard spans HPD violations, immediately-hazardous (Class C) conditions, ECB/OATH fines, emergency repair orders, facade (LL11) status, sidewalk-shed persistence, LL97 and LL84 compliance, AEP designation, HPD-initiated litigation, modeled reserve-fund adequacy, vacate orders, and rent-stabilization exposure — each normalized to a common 0-100 severity scale so that 50 means “meaningfully elevated” and 80+ means “severe” in every dimension. The per-dimension calibration constants are part of the proprietary model.
EXTRACTION
The four-bucket spending classifier.
Every post-occupancy DOB filing at a building is classified into one of four buckets based on whether the work was required by law, added value, or is unexplained by either:
REQ — Required by law
Work mandated by statute: LL11 facade inspections, LL97 emissions compliance, sprinkler/standpipe upgrades, fire-alarm replacements, life-safety work. Not extraction. This spending is non-discretionary.
VAD — Value-add reinvestment
Capital improvements that benefit residents and building value: lobby renovations, HVAC upgrades, roof replacements, elevator modernization, commercial-tenant buildouts. Not extraction. This spending creates or preserves value.
DSC — Discretionary opaque
Post-occupancy spending that cannot be attributed to a legal mandate or a clear reinvestment. The residual category. These are the filings where the purpose is not self-evident from the DOB record.
FINES — ECB/OATH fines + HPD ERP
Enforcement costs: fines imposed by the city for violations, plus emergency repair costs billed to the building by HPD. These represent governance failure costs, not investment.
The headline extraction figure in each audit report is discretionary-opaque + fines/ERP: the portion of post-occupancy spending that cannot be explained by a legal mandate or a clear reinvestment in the building. This is not an accusation of fraud; it is a measurement of spending that deserves scrutiny.
LIMITATIONS
What the grade does not measure.
- Interior apartment condition. HPD violations and DOB filings describe common-area and building-system conditions. They do not inspect individual apartments unless a complaint is filed.
- Management quality beyond public record. A building with a competent managing agent who resolves issues before they become violations will score better. That is by design. But responsiveness, communication quality, and resident satisfaction are not measured.
- Financial health. Reserve-fund adequacy is modeled from visible capital spend, not from audited financial statements (which are not public in New York). The model is a floor estimate, not a precise measurement.
- Board governance quality. Whether board meetings are held, whether minutes are distributed, whether conflicts of interest are managed. None of this is captured in any public dataset.
- Market value or investment potential. The grade measures governance and compliance, not price trajectory or rental yield.
BIAS PREVENTION
How the system avoids bias.
No neighborhood weighting
A building in the South Bronx is graded with the same formula as a building on Park Avenue. There is no adjustment for neighborhood, borough, median income, or property value.
No ownership-type adjustment
Condos and co-ops are graded identically. Sponsor-controlled buildings and resident-controlled buildings face the same thresholds.
No editorial override
No person at CondosCoopsNYC or TASFGA can manually change a building's grade. The grade is computed by code from data. The code is the same for every building.
Dismissal-aware
When 100% of a building's DOB high-severity violations have been dismissed on appeal, the system records that as context rather than a grade trigger. Issuance volume is noted, but a clean dismissal is not penalized.
Absence is a finding
If a dataset returns zero records for a building, that absence is documented in the Data-Source Coverage Ledger. It is not hidden. A building that doesn't appear in a dataset might be clean, or the dataset might not cover it. The report says which.
No building-age penalty
A 1920s co-op is graded on its current violation record, not on its age. Older buildings are not penalized for being older. If an older building has more violations, those violations are what move the grade, not the year it was built. Age-cohort comparisons appear in Level 3 reports as context, but they do not affect the letter.
See the grades in action.
Browse the full graded catalog or request an audit report for a specific building.