Findings №4 · New Haven, CT · June 2026

The second boom has new names.

In 2025, corporate buyers closed on 712 New Haven parcels — within four percent of the all-time peak set in 2022. But this is not a rerun. Mandy, who drove a third of the 2021 wave, took just 10% of this one. More than half of the 2025-26 buying traces to entities our clustering has never been able to link to any known group — and 60% of it came from LLCs formed in 2024 or later. A new cohort is assembling portfolios, and the public record is meeting most of them for the first time.

§1The wave came back

Quiet for two years. Then 2025.

After the 2019-2022 boom crested at 741 acquisitions, the pace cooled into the low five-hundreds — elevated by historical standards, but trending down. Then 2025 broke the trend in the other direction: 712 corporate acquisitions, the second-highest year ever recorded in this dataset, within four percent of the peak. The first months of 2026 are tracking a similar pace.

2019360
2020460
2021675
2022741
2023526
2024510
2025712
2026 YTD170

The first boom had an obvious protagonist: Findings №1 documented Mandy closing a deal every 37 hours through 2021. The second boom's most interesting feature is that it doesn't. No single group dominates it. The buying is broader, newer, and harder to attribute — which is precisely what makes it worth watching.

§2Who's buying now

The incumbents stepped back. Someone else stepped in.

10%
Mandy's share of 2025-26 buying (was 35% of 2021)
53%
of 2025-26 purchases by entities linked to no known group
60%
of 2025-26 purchases by LLCs formed in 2024 or later

During the first boom, four out of five purchases traced to an established ownership network — a group with a history of filings, principals, and addresses our clustering could resolve. In 2025-26 that flipped: 53% of acquisitions belong to no known group. That's not because the buyers got better at hiding. It's mostly because they're new — three of every five 2025-26 purchases were made by an LLC whose Connecticut filing is less than two years old. There's no graph to connect them to yet.

  1. 1Mandy85 parcels
  2. 2Menahem Edelkopf49
  3. 3KATAN HOMES LLC (no group — see §3)36
  4. 4Hecht14
  5. 5Ocean13
  6. 6Dina Landau · Yidah Blum · Townsend Ave Development10 each

The established names didn't vanish — Mandy and Edelkopf still top the table. But their combined 134 parcels are a sixth of the wave, not half of it. The rest is a long tail of names the Secretary of State met for the first time in 2024 and 2025.

"The first boom consolidated the city into known hands. The second is minting new ones."
§3The institutional handoff

A fund exits. Thirty-six houses change hands in one quarter.

The single largest unattributed buyer of the second boom is KATAN HOMES LLC. In early 2026, over roughly one quarter, it took title to 36 New Haven parcels — $8.3M in assessed value, mostly single-families and three-families — every one of them from the same seller: REF II SFR 1 DE LLC, a Delaware-filed single-family-rental fund vehicle that had assembled the portfolio during the first boom.

Every deed in the handoff recorded at $0 — a portfolio-level transaction whose price lives in the fund's books, not the city's land records. And the buyer surfaces nothing through the Connecticut Secretary of State's standard business search, the same out-of-state opacity pattern Findings №2 documented for LLPs.

One more wrinkle the filings add: the selling fund is not a stranger to this map. REF II SFR 1 DE LLC's registered agent is Mandy Management LLC, its business address is 399 Whalley Avenue, and our clustering attributes it to the Mandy network. So this is either New Haven's largest landlord shedding a 36-house portfolio to an outside buyer, or an internal re-titling into a fresh vehicle — and a $0 deed cannot tell you which. That ambiguity is the finding.

Read together, that's a story the parcel record tells almost on its own: a fund vehicle assembled during the cheap-money years has handed off three dozen homes — with the public record showing who signed, but not for how much, and not yet for whom.

§4The flip economy

Bought at breakfast, sold by dinner.

The second boom is also the busiest flipping market in the dataset. We count 476 parcels resold within twelve months of purchase (both transactions arms-length priced) across the full record — median markup $36,000 — and 2025 produced more of them (27) than any other year. The sharpest subset: 148 same-day double closings, where an entity took title and conveyed it within one recording day, for a combined paper gain of $2.9M.

  1. 1437 Dixwell Ave — OREO NH LLC, Nov 2024$190K → $455K+$265K same day
  2. 2543 Woodward Ave — Zohar LLC, Jun 2023$1.0M → $1.25M+$250K same day
  3. 3680 Dixwell Ave — OREO NH LLC, Oct 2023$200K → $365K+$165K same day

Same-day double closings are the signature of wholesaling: contract a property below market, find the end buyer, run both deeds through the land records together. It's legal, and it's a measure of how much spread sits between what a distressed or under-marketed property fetches and what an investor will pay an hour later. The most frequent sub-year flipper in the dataset, SUN ESCAPE LLC, has run the play seventeen times.

Underneath the flipping is the price floor moving. The median arms-length three-family — the city's workhorse rental building — sold for $262,500 in 2019, $299,900 in 2021, and $400,000 in 2025: up 52% over six years. Whoever is buying in the second boom is paying first-boom-plus prices, which says something about what they expect rents to do.

§5What we'll watch

Open questions the next refresh starts answering.

Sources & Methodology

  1. New Haven VGSI assessor records — sale dates, prices, and full per-parcel deed chains (book/page-ordered). Acquisition years use each parcel's most recent sale date; 2025-26 cohort = parcels whose current owner took title in 2025 or 2026.
  2. CT Secretary of State business filings — LLC formation dates ("formed 2024 or later" uses the dateFormed on file for the acquiring entity). Entities with no CT filing surfaced (e.g. Katan Homes LLC) are noted as such.
  3. Flip detection: consecutive deed-chain entries on one parcel where both transfers carry arms-length prices (>$0) and different owners, within 365 days (or the same recording day for double closings). Nominal-price intra-network re-deeds are excluded by construction.
  4. Three-family price medians use arms-length sales above $50,000 in the stated calendar year, restricted to the "Three Family" assessor use class to avoid mix effects.

All figures are a June 2026 snapshot of a weekly-refreshed dataset. The unattributed share is a floor on attribution, not a claim of concealment: a new LLC with one filing and one parcel gives a clustering algorithm nothing to link. Some of the 2025-26 cohort will resolve into known or new networks as filings accumulate — that's the point of watching.