thesisPrototype

Government data

US New Home Sales SAAR, July 2026

What will the U.S. Census Bureau and HUD first report for sales of new single-family houses in July 2026, seasonally adjusted annual rate, in thousands?

Forecast

620k
538k–702k
80% prediction interval
gpt-5.5Run record ↗

Cumulative probability

Chance that the outcome is at or below each value.

0%25%50%75%100%415k620k825kForecast value

Shaded band: 80% interval (538k–702k). Dashed line: point estimate (620k).

Derived from the point estimate and 80% interval; the agent did not report a full distribution.

Observed outcomeinside 80% interval
actual
607k
forecast
620k with 80% interval [538k, 702k]
error
-13k · absolute 13k
cdf score
CRPS 18.6 · PIT 0.44
source
census_new_home_sales New Residential Sales, Table 1

First print for 2026-07 captured from https://alfred.stlouisfed.org/graph/alfredgraph.csv?id=HSN1F&vintage_date=2026-08-25 on the official release date named by the cell's resolver.

Analysis

§US July 2026 New Single-Family Home Sales SAAR

Framing and exact resolver: the target is Census/HUD New Residential Sales Table 1a, United States new single-family houses sold, seasonally adjusted annual rate, in thousands, first print for July 2026. The Census release schedule verifies the July 2026 New Residential Sales release for August 25, 2026 at 10:00 AM; the June release also says the July report is scheduled for August 25, 2026.

▸ Reported tool use: official.lookupmodel report
Read Census New Residential Sales current release for latest headline and revision context.
↳ Fetched New Residential Sales June 2026: sales 628,000 SAAR; May 2026 rate 618,000; June 2025 rate 665,000; for-sale inventory 485,000; months' supply 9.3.
▸ Reported tool use: official.lookupmodel report
Read Census New Residential Sales PDF Table 1a for recent seasonally adjusted history in the same variant.
↳ Fetched Table 1a United States SAAR values, thousands: 2025-06 665, 2025-07 648, 2025-08 698, 2025-09 714, 2025-10 652, 2025-11 757, 2025-12 723, 2026-01 576, 2026-02 630, 2026-03 659, 2026-04 646, 2026-05 618, 2026-06 628.
▸ Reported tool use: official.lookupmodel report
Read Census New Residential Construction current release for leading construction indicators.
↳ Fetched June 2026 construction indicators: total permits 1,367,000 SAAR, revised May permits 1,410,000, single-family authorizations 871,000, revised May single-family authorizations 892,000, single-family starts 895,000.

Base rate/reference class: recent same-variant monthly SAAR changes are very noisy. A persistence prior from the latest 628 thousand is the natural base rate, with recent changes of -17, +50, +16, -62, +105, -34, -147, +54, +29, -13, -28, and +10 thousand across the fetched 2025-06 to 2026-06 sequence.

Level and momentum: the latest print improved only 10 thousand from May and remained 37 thousand below June 2025. The May-to-June rebound argues against extrapolating the May weakness, but the year-over-year decline and high 9.3 months' supply argue against a strong July acceleration.

Mechanism adjustment: June single-family permits fell to 871 thousand from 892 thousand, and total permits fell to 1,367 thousand from 1,410 thousand. That does not mechanically determine sales, but it supports a small downward adjustment from pure 628 persistence rather than an upward continuation.

Prior/update/interval: persistence prior = 628. Adjustment components: -5 for soft single-family permits, -3 for elevated 9.3 months' supply and below-year-ago sales, giving point = 628 - 8 = 620. Historical sample = same-variant monthly SAAR changes from 2025-06 through 2026-06. The sample standard deviation of successive changes is sigma = 64.25 thousand. 80% half-width = 1.28*sigma = 1.28*64.25 = 82.24 thousand, rounded to 82. Interval = 620 - 82 to 620 + 82 = 538 to 702.

Counter-consideration: upside risk is a July demand rebound from rate relief or builder incentives that would land above the interval, above 702 thousand, especially if the South recovers from June weakness. Downside risk is a renewed sales drop from high mortgage rates, cancellations, or excess inventory that would land below the interval, below 538 thousand. Values outside the interval are plausible because this series has large sampling and month-to-month volatility.

Review disposition: accepted the reviewer suggestion to make tail thresholds explicit, identifying upside above 702 and downside below 538. The official Census schedule URL was already included in sourceContext and the resolver text already names the verified August 25, 2026 release date, so no resolver change was needed.

Key drivers

  • June sales rebounded modestly from revised May but stayed below June 2025
  • High months' supply and elevated for-sale inventory lean against a large upside move
  • Single-family permits softened in June, pointing to cautious builder conditions
  • Large month-to-month noise in new-home sales keeps the interval wide

Tool evidence

This run has no captured tool responses. “Reported tool use” in the analysis is the model’s account, not an archived tool response.

Sources and resolution

Official source
U.S. Census Bureau and HUD New Residential Sales, Table 1, first print archived for HSN1F
Resolved
August 25, 2026
Resolution rule
Resolve to the first published July 2026 value for New Houses Sold, United States, seasonally adjusted annual rate, Table 1a/New Residential Sales series HSN1F, in thousands. Use the first official print only; ignore later revisions and seasonal-revision changes unless the original release is formally withdrawn before publication.
Run details

The analysis is the model’s written report. Tool-use descriptions in that report are model claims; the activity artifacts contain the execution record.

thesis.analyst · gpt-5.5 · fast · v2.3.0

pre-submit review · completed

The draft is publishable: resolver, prior/update logic, volatility-based interval, tail scenarios, and JSON fields are coherent with the target contract.

  • info optional_suggestion: Name the numeric tail thresholds explicitly, e.g. upside above 702 and downside below 538.
  • info optional_suggestion: Consider adding the official Census schedule URL directly to the resolver/source text alongside the ALFRED first-print archive.

disposition not applicable: Review disposition: accepted the reviewer suggestion to make tail thresholds explicit, identifying upside above 702 and downside below 538. The official Census schedule URL was already included in sourceContext and the resolver text already names the verified August 25, 2026 release date, so no resolver change was needed.

disposition not applicable: Review disposition: accepted the reviewer suggestion to make tail thresholds explicit, identifying upside above 702 and downside below 538. The official Census schedule URL was already included in sourceContext and the resolver text already names the verified August 25, 2026 release date, so no resolver change was needed.

Activity artifacts

Complete original trace

§US July 2026 New Single-Family Home Sales SAAR

Framing and exact resolver: the target is Census/HUD New Residential Sales Table 1a, United States new single-family houses sold, seasonally adjusted annual rate, in thousands, first print for July 2026. The Census release schedule verifies the July 2026 New Residential Sales release for August 25, 2026 at 10:00 AM; the June release also says the July report is scheduled for August 25, 2026.

▸ Reported tool use: official.lookupmodel report
Read Census New Residential Sales current release for latest headline and revision context.
↳ Fetched New Residential Sales June 2026: sales 628,000 SAAR; May 2026 rate 618,000; June 2025 rate 665,000; for-sale inventory 485,000; months' supply 9.3.
▸ Reported tool use: official.lookupmodel report
Read Census New Residential Sales PDF Table 1a for recent seasonally adjusted history in the same variant.
↳ Fetched Table 1a United States SAAR values, thousands: 2025-06 665, 2025-07 648, 2025-08 698, 2025-09 714, 2025-10 652, 2025-11 757, 2025-12 723, 2026-01 576, 2026-02 630, 2026-03 659, 2026-04 646, 2026-05 618, 2026-06 628.
▸ Reported tool use: official.lookupmodel report
Read Census New Residential Construction current release for leading construction indicators.
↳ Fetched June 2026 construction indicators: total permits 1,367,000 SAAR, revised May permits 1,410,000, single-family authorizations 871,000, revised May single-family authorizations 892,000, single-family starts 895,000.

Base rate/reference class: recent same-variant monthly SAAR changes are very noisy. A persistence prior from the latest 628 thousand is the natural base rate, with recent changes of -17, +50, +16, -62, +105, -34, -147, +54, +29, -13, -28, and +10 thousand across the fetched 2025-06 to 2026-06 sequence.

Level and momentum: the latest print improved only 10 thousand from May and remained 37 thousand below June 2025. The May-to-June rebound argues against extrapolating the May weakness, but the year-over-year decline and high 9.3 months' supply argue against a strong July acceleration.

Mechanism adjustment: June single-family permits fell to 871 thousand from 892 thousand, and total permits fell to 1,367 thousand from 1,410 thousand. That does not mechanically determine sales, but it supports a small downward adjustment from pure 628 persistence rather than an upward continuation.

Prior/update/interval: persistence prior = 628. Adjustment components: -5 for soft single-family permits, -3 for elevated 9.3 months' supply and below-year-ago sales, giving point = 628 - 8 = 620. Historical sample = same-variant monthly SAAR changes from 2025-06 through 2026-06. The sample standard deviation of successive changes is sigma = 64.25 thousand. 80% half-width = 1.28*sigma = 1.28*64.25 = 82.24 thousand, rounded to 82. Interval = 620 - 82 to 620 + 82 = 538 to 702.

Counter-consideration: upside risk is a July demand rebound from rate relief or builder incentives that would land above the interval, above 702 thousand, especially if the South recovers from June weakness. Downside risk is a renewed sales drop from high mortgage rates, cancellations, or excess inventory that would land below the interval, below 538 thousand. Values outside the interval are plausible because this series has large sampling and month-to-month volatility.

Review disposition: accepted the reviewer suggestion to make tail thresholds explicit, identifying upside above 702 and downside below 538. The official Census schedule URL was already included in sourceContext and the resolver text already names the verified August 25, 2026 release date, so no resolver change was needed.

calibrated forecast · 80% CI
620k[538k · 702k]
Target metadata

Data point: census.new_residential_sales.new_single_family_houses_sold_saar.2026_07.first_print

More government data forecasts