So I've spent the last few months cross-referencing 22+ NYC public data sources (HPD violations, ECB fines, tax liens, litigation, 311 complaints, ACRIS mortgages, DOB permits, energy benchmarking, etc. etc.) to score every multifamily building in the city by distress level. The dataset covers 172,000+ buildings and roughly 37 million individual records. Figured I'd share some patterns that surprised me.

Numbers

As of this week there are about 242 buildings in what I'd call Critical condition (in the worst shape) and 4,494 in High distress. That sounds like not much with 172K buildings, but these are concentrated in certain neighborhoods and portfolios. The Critical count has roughly tripled over the past 12 months. 82 last April, peaked at 264 in early March. There's been a slight pullback now that heating season is ending, but the structural trend is still up.

Heat complaints!!

Winter heat complaints are clearly the single strongest predictor of future distress. When a building starts getting 311 "no heat/hot water" calls in November, HPD violations follow about 30 days later. Litigation follows about 90 days after that. By spring the building is in a completely different financial position than it was in October. If you're looking for acquisition targets, looks like tracking heat complaints during winter gives you a few months' head start before anything shows up in the usual places.

That concentration

The Bronx leads by a wide margin. Both absolute count and per-capita. Specifically the 167th-180th Street corridor and the Grand Concourse area. Brooklyn (Flatbush, Bed-Stuy, Crown Heights) is second. Manhattan has fewer distressed buildings overall, but the ones that are distressed tend to be larger. 100+ unit buildings with massive violation backlogs. 79 portfolio landlords own 3 or more distressed buildings each. These aren't one-off situations. Same ownership, same deferred maintenance playbook, across multiple properties. Some of these portfolios have $500K+ in outstanding ECB fines across their buildings.

Large-building blind spot?

This one caught me off guard. I trained an ML model (XGBoost) on historical sales to predict which buildings are heading toward distress, and it picked up something an initial heuristic scoring tool I built had missed: large buildings with 100+ open violations score LOW on standard metrics because everything gets divided by unit count. Like a 1,000-unit tower with 200 open violations looks "fine" at 0.2 violations per unit. But 200 open violations in any building is a crisis. The model learned to weight absolute violation counts alongside per-unit rates, and it flagged dozens of large buildings that traditional analysis completely misses. Some of these are well-known addresses.

>>Safer Homes Act implications

7,773 buildings currently meet the thresholds that could trigger action under the Safer Homes Act. That's not a prediction. It's just applying the statute's criteria to current public data. Whether enforcement actually follows is a different question, but if you own or are looking at a building near those thresholds, worth knowing what you're looking at.

What next then

Buildings where ECB fines are growing, mortgage LTV is above 1.0, and there's been zero permit activity in 24 months. That combo tends to show up right before a distressed sale. There are about 400 buildings in that bucket right now. The underlying data is all public record, so even if you're pulling your own HPD/ECB data, the heat-complaint-to-litigation pipeline and the large-building blind spot are worth looking for in your own analysis.