Dwellsy TrendsIQ

Overview

Dwellsy TrendsIQ is a dataset of U.S. rental housing market trends built on first-party data from over 17 million professionally managed rental units. The dataset is sourced from the systems of record of over 25,000 property managers across the United States through real-time, direct integrations. TrendsIQ captures rents at the point where pricing decisions are made and transforms raw rental records into aggregated market trend statistics.

Under Dwellsy's v2 methodology, each ZIP code's contribution to city- and MSA-level figures is weighted using verified rental inventory (URU, "Unique Rentable Unit," counts) combined with Census renter-occupancy data, rather than being weighted equally across ZIPs. Rent trends are computed separately across six independent strata by property type and bedroom count, using a 3-month rolling comparison window. Machine-learning models extend estimated coverage into data-sparse and less urbanized markets, with each figure labeled to indicate whether it reflects direct data or a model estimate. See Methodology below for details, including the differences between the v1 and v2 methodologies.

Data Description

The dataset provides aggregated statistics of rental market trends at three geographic levels: MSA (Metropolitan Statistical Area), ZIP code, and city, delivered as three separate tables. The underlying data originates from over 17 million professionally managed rentals, with 30+ integrations across property management systems and real-time data from 25,000+ property managers.

Within each geography, monthly rent figures are tracked separately across six independent strata: apartment rentals with 0, 1, or 2 bedrooms, and single-family rentals with 2, 3, or 4 bedrooms. The observation level consists of one set of statistics per geography, per month, per stratum.

Field dictionary

Field	Description
Id	Unique identifier assigned to each aggregated record in the dataset.
MSA	The Metropolitan Statistical Area where the property data belongs.
City	Name of the city where the data is aggregated.
ZIP Code	ZIP code representing the geographic boundary for the aggregated data.
Bedrooms	0, 1, and 2 for apartments; 2, 3, and 4 for single-family rentals.
Address Type	Indicates whether the property is an apartment or a single-family rental.
Month	The month in which the rent data was recorded.
Rent Amount	The median rent price, calculated after a data-cleaning and smoothing process that includes Gaussian filtering, LOESS smoothing, and additional statistical methods.
Rate of Change	The percentage change in median rent compared to the previous year for the same unit.

Coverage

Geographic Coverage: The dataset covers rental markets across the United States at three levels of geographic aggregation: MSA, ZIP code, and city.

Universe: The dataset encompasses over 17 million professionally managed rental units sourced from over 25,000 property managers.

Methodology

The dataset is not scraped from listing sites and does not rely on voluntary surveys. TrendsIQ is built through the following pipeline:

Data Sources — The pipeline starts from the Dwellsy IQ dataset of first-party rental listings.
Data Cleaning — Duplicate and incorrect listings are removed.
Data Correction — Temporary distortions and anomalies are adjusted.
Data Smoothing — Exponential moving average (EMA), LOESS, and Gaussian methods reduce noise across a 3-month rolling window.
Hybrid-Weighted Aggregation — ZIP-level medians are aggregated to city and MSA levels using weights based on verified rental inventory (URU counts) and Census renter-occupancy data.
Machine Learning — Models extend coverage into data-sparse markets without relying on a spurious year-over-year comparison.

Hybrid ZIP weighting. Each ZIP's influence on city and MSA figures is tied to its share of verified rental inventory rather than to listing frequency: a ZIP representing 2% of an MSA's rental stock contributes 2% of the weight for that MSA's figure.

Unique Rentable Units (URUs). The verified rental inventory used for weighting is counted in URUs — Dwellsy's permanent unit-level identifiers. Every rentable asset (house, apartment unit, duplex, townhouse, mobile home, room, bed, storage, or parking space) is assigned a single permanent ID when it first enters the Dwellsy IQ dataset; records arriving from multiple sources are reconciled and matched against existing records, and a new ID is issued only when no URU exists for the unit. The ID remains unchanged even if an address is reformatted or a street is renamed. Because inventory is counted at the unit level rather than the listing level, duplicate and relisted records do not inflate a ZIP's weight. Unit-level URU fields (uru_id, uru_type) are delivered in the Dwellsy TotalIQ listings dataset.

Six independent strata. Apartment and single-family rent trends are measured separately by bedroom count (apartments: 0, 1, 2 bedrooms; single-family: 2, 3, 4 bedrooms) to avoid cross-property-type distortion.

Responsiveness window. A rolling 3-month comparison is used to detect market shifts, which Dwellsy reports as detecting turning points roughly six weeks sooner than the prior 6-month window, while maintaining year-over-year comparability.

Tiered coverage framework. Each figure is labeled according to its source — direct data, machine-learning estimate, or new-market coverage — so that direct observations and modeled estimates are distinguishable.

Methodology versions

Dwellsy revised how TrendsIQ aggregates and weights rent data when rolling up from individual listings to ZIP, city, and MSA-level figures. Per Dwellsy, the underlying listing data itself is unchanged between versions; only the weighting and aggregation methodology changed.

Feature	V1	V2
ZIP weighting	Equal ZIP weighting	Weighted by URU + Census
Rolling window	6 months	3 months
Reporting lag	Up to 3 months	Roughly 1.5 months
Turning-point detection	Slower to detect shifts	Roughly 6 weeks faster
Stability in thin-data ZIPs	Sparse ZIPs add noise	Sparse ZIPs have less influence

A note on methodology timing. Periods delivered through the May-2026 data month were computed under Dwellsy's v1 methodology. Deliveries from the June-2026 data month onward use v2. Dwellsy states that the underlying listing data is unchanged between versions — only the weighting and aggregation logic changed. Whether previously delivered history is restated under v2 depends on whether Dwellsy re-delivers that history to Dewey; the Dewey panel stores values as they were delivered and does not itself restate prior periods.