The Address Index
Greenwich · Single-family

Greenwich real estate, measured.

A stable view of Greenwich, Connecticut home prices, housing inventory, sales velocity, and market heat—built from public, source-linked records without letting one unusual month define the trend.

Address Index106.2+0.4% YoY · 2024 = 100 · nowcast
12-month market level$3.26Mstable headline signal
Latest monthly median$3.60Mmix-sensitive context
Active inventory113homes
Market heat53/100Balanced
Updated monthly

Greenwich Market Dashboard

Through July 2026
Greenwich single-family dashboard with the quality-adjusted Address Index from 2001 through July 2026: latest index 106.2, 12-month market level $3.26M, 113 active listings, and 36 average days on market.

The main view now runs from 2001 to the present. The solid green line is the quality-adjusted Address Index built from repeat sales of the same properties; the dashed gold ending is a transparent nowcast after complete transaction coverage. “Recent” keeps a full year of fitted history visible before the nowcast; use 5Y for a wider current-cycle view.

Explore the Greenwich Home Price Index from 2001 to today →

Public-record intelligence

Synthetic Twin

Choose a completed sale to inspect its out-of-sample model expectation, a tighter likely zone, broader uncertainty, the public facts behind it, and where its earlier twins sit across Greenwich.

Synthetic Twin chart for 147 Old Church Road, which sold for $2.80M on August 26, 2025. The public-record model expected $4.18M, with a 50% likely zone of $3.62M to $4.83M and a 80% broader range of $2.89M to $6.05M. A Greenwich map locates the sale and its earlier public-sale twins.

Synthetic Twin · 147 Old Church Road
147 Old Church Road sold for $2.80M, 33.0% below its $4.18M out-of-sample expectation. Its 50% likely zone is $3.62M–$4.83M; the 80% broader range is $2.89M–$6.05M. The model used 4,813 training, 1,054 calibration, and 1,055 untouched test sales; test median error was 16.5%. This is a market comparison, not an appraisal.

The dashboard is open during launch

Get the judgment, not just the charts.

Explore the live Greenwich dashboard freely. Paid Address Index membership delivers the complete weekly analysis, adjusted comparable-sale work, model diagnostics, monthly State of Greenwich deep dives, and the paid archive—all through Substack, with no second login.

The chart library

Price, leverage, liquidity, and the long view

Composite signal

Market heat

53 / 100

Balanced

A relative score of price momentum, inventory tightness, market speed, and trailing sales liquidity.

What drives it

Heat score components

Inventory Tightness
78
Liquidity
64
Market Speed
96
Price Momentum
1
Recent history

Monthly tape

MonthReported median12-month signalMarket-level indexClosingsInventorySupplyLeverageDays
Jul 2026$3.60M$3.26M117.7741132.6 mo5336
Jun 2026$3.81M$3.25M117.4581222.8 mo5237
May 2026$3.15M$3.26M117.7501142.6 mo4869
Apr 2026$4.00M$3.22M116.336992.2 mo6139
Mar 2026$2.79M$3.16M114.029892.0 mo5385
Feb 2026$2.57M$3.28M118.321791.8 mo6657
Jan 2026$4.49M$3.38M122.037741.7 mo7191
Dec 2025$3.15M$3.34M120.647561.3 mo7665
Nov 2025$3.06M$3.32M119.844962.2 mo7669
Oct 2025$2.98M$3.25M117.4361202.8 mo6083
Sep 2025$2.70M$3.29M118.6311192.8 mo8031
Aug 2025$2.80M$3.34M120.5581092.5 mo8255
Evidence ledger

Latest Address Index quarters

QuarterIndex95% intervalEvidenceRepeat-pair evidence
2026 Q2106.290.6–124.7nowcast0
2026 Q1103.290.0–118.3nowcast0
2025 Q4109.197.3–122.3nowcast0
2025 Q3107.4102.2–112.8observed94
2025 Q2105.8101.0–110.8observed84
2025 Q1102.398.2–106.6observed65
2024 Q4102.099.3–104.6observed72
2024 Q3100.498.9–101.9observed117
2024 Q299.397.8–100.8observed99
2024 Q198.496.1–100.7observed61
2023 Q492.688.9–96.3observed67
2023 Q390.286.0–94.6observed115
Transparent by design

Method and data quality

Address Index

A weighted repeat-sales model of the same Greenwich single-family properties over time. Quarterly estimates are smoothed over four quarters and scaled so the 2024 average equals 100.

Same House Test

Splits repeat-sale intervals into 1,568 with public evidence of rebuilding or substantial modernization and 3,824 with no recorded major change. It is an evidence comparison, not an estimate of renovation ROI.

Synthetic Twin

The data stays chronological: 4,813 earlier sales fit the model, 1,054 later sales set the ranges, and 1,055 newest sales remain untouched for the final test. Median test error is 16.5%. The 50% likely zone captured 51.3% of test sales; the 80% broader range captured 82.1%. Market comparison, not an appraisal.

Observed, reconstructed, nowcast

Observed quarters have direct state transaction coverage. Reconstructed quarters rely more heavily on public parcel sale histories or interpolation. After 2025 Q3, the dashed line is a transparent nowcast from official monthly reports.

Monthly dots

Each gold dot is the official median for one month. The dots are not connected because they describe different baskets of sold homes and should not be read as month-to-month appreciation.

Time on market

The long-run line is a listing-count-weighted composite of Realtor.com median listing time across Greenwich's five ZIP codes since 2016. The recent GAR line is the reported average for sold single-family homes. They are labeled separately because they are not the same statistic.

Buyer–Seller Leverage

Applies the published Market Heat weights to every complete month. Below 35 is buyer-leaning, 35–65 balanced, and above 65 seller-leaning. It is relative to the displayed Greenwich history, not a promise about a particular negotiation.

Inventory Runway

Active single-family inventory divided by the monthly closing pace implied by the trailing 12 months. It measures buyer choice at the recent pace—not how long a particular listing will take to sell.

The Luxury Mix

Groups 17,610 deduplicated public sales into four nominal price bands and shows each band’s trailing four-quarter share. The latest available transaction quarter is labeled separately because it can lag monthly reports. Mix is not appreciation.

Model Surprise Map

Plots verified coordinates for untouched Synthetic Twin test sales. Green cleared more than 10% above the chronological public-record expectation, gold more than 10% below, and gray within ±10%. It shows where the public model missed—not why.

Uncertainty

The shaded range is a 95% statistical interval. It widens when direct observations are sparse and widens further for the current nowcast. The center line is an estimate, not an appraisal.

Current coverage

20,159 state transaction rows, 4,842 Town archive rows, 19,364 parcel records, and 5,538 eligible repeat-sale pairs.

Sources and audit

State transactions · Town assessor archive · Town GIS map · listing-time history · latest market report · validation JSON.

Directional market intelligence only—not an appraisal, brokerage opinion, or investment recommendation.