Understand the market.
Model the build.
Construction inputs, new-home demand and development economics in one working view. Follow the evidence from a material cost trend to the sale price a project needs.
Observation frequency: monthly. Last successful retrieval: Oct 12, 2026, 3:45 PM. These are producer indexes and national housing measures.
Shared source cache is checked again after Oct 13, 2026, 7:54 PM when the engine is used.
01 / Construction cost inputs
See which material categories are moving.
Monthly U.S. producer price indexes track changes in selling prices received by producers. They provide an escalation signal. A delivered purchase price requires an item specification, quantity, supplier and delivery location.
| Material | Observation | Index level | 1-month change | 12-month change | Evidence |
|---|---|---|---|---|---|
| Index 1982=100 | Aug 2026 | 294.899 | -1.9% | +11.8% | Source ↗ |
| Index 1982=100 | Aug 2026 | 395.186 | +0.1% | +2.3% | Source ↗ |
| Index 1982=100 | Aug 2026 | 355.665 | 0.0% | -1.2% | Source ↗ |
| Index 1982=100 | Aug 2026 | 484.844 | +0.7% | +0.2% | Source ↗ |
| Index 1982=100 | Aug 2026 | 381.162 | +1.7% | +23.4% | Source ↗ |
| Index Dec 1986=100 | Aug 2026 | 602.48 | +4.2% | +27.2% | Source ↗ |
| Index 1982=100 | Aug 2026 | 374.098 | +0.5% | +3.9% | Source ↗ |
| Index 1982=100 | Aug 2026 | 272.374 | +1.0% | -0.9% | Source ↗ |
A national softwood lumber category; not a local grade/species/length quote or lumber futures dollars per thousand board feet.
02 / Experimental forecast lab
A forecast you can audit.
Inspect the observed history, forecast horizon and separate historical test. The model forecasts the selected national series; it does not forecast a supplier quote or a subdivision’s sales.
Softwood lumber
Index 1982=100Not seasonally adjusted · United States · observed history from 2015
Only observed values are drawn. Missing observations stay missing; the forecast endpoint is shown separately below.
Damped 12-month log trend
The selected model did not beat the last-value baseline in the separate historical validation. Treat this experimental forecast cautiously.
Errors use the selected series’ units. Backtests use currently revised history, not unrevised releases available at each past date. This is an experimental statistical model, not a record of live forecast performance.
Translate the scenario into cost exposure.
Enter a material-package budget stated at Aug 2026 prices. Apply the index change only to the uncommitted share. Keep labor, locked contracts, taxes and freight separate. This produces a budget scenario, not a new quote.
Enter a valid budget
Uses the selected model scenario
Model selection, validation and limits
Three candidates compete: no change, seasonal naive (same calendar month last year), and a damped log trend estimated from the last 12 months. Earlier rolling forecast origins select the lowest median absolute percentage error. Twelve later origins assess it separately, with a horizon-dependent gap so selection outcomes are known before validation begins.
The forecast is refitted to the full available history. The displayed range applies the 10th and 90th percentiles of 12 historical log forecast errors to the endpoint. This small-sample range has no promised coverage. Historical validation windows overlap; results can be sensitive to the cycle and revisions. A larger AI model has not been deployed behind this output.
Foundation-model candidates such as Chronos-2 ↗ require a separate runtime and evaluation on these exact targets before use. More complex models do not automatically improve the forecast.
Rolling-origin evaluation ↗03 / For-sale new construction
Track demand for newly built homes.
Aug 2026 · United States
U.S. new single-family home sales
684Thousands of homes, annual rate · Seasonally adjusted annual rateNational new one-family houses sold, expressed in thousands at an annualized monthly pace. Multiply by 1,000 to read the annual rate. This is not the number sold in a single month or a local subdivision absorption forecast.
Aug 2026 · United States
U.S. new-home months of supply
8.5Months of supply · Seasonally adjustedNational ratio of new houses for sale to the current new-house sales pace. It is not days on market, completed-home-only inventory, or a forecast of how long a particular home will take to sell.
New-home sales are recorded at contract/deposit, which may precede construction. National sales and inventory are context; local community sales, cancellations, incentives and competing lots still need direct evidence.
04 / Local construction context
Start with the metro. Verify the project.
Six full metropolitan areas, reviewed September 5, 2026. These local figures are a fixed reviewed snapshot; the connected national feed above refreshes separately. Single-unit authorizations include for-sale and rental homes.
| Metro | Single-unit authorized | vs Jan–Jul ’25 | Payroll YoY | Broad HPI YoY | Sources |
|---|---|---|---|---|---|
| Charlotte | 7,871 | -27.6% | +1.2% | +1.5% | Permits · readable guide → · Jobs ↗ · HPI · readable guide → |
| Raleigh | 7,405 | +2.8% | +2.3% | +1.1% | Permits · readable guide → · Jobs ↗ · HPI · readable guide → |
| Dallas–Fort Worth | 24,983 | -3.3% | +1.1% | +0.6% | Permits · readable guide → · Jobs ↗ · HPI · readable guide → |
| Houston | 28,302 | -5.0% | +1.2% | +3.4% | Permits · readable guide → · Jobs ↗ · HPI · readable guide → |
| Phoenix | 13,827 | -12.0% | +0.9% | +0.8% | Permits · readable guide → · Jobs ↗ · HPI · readable guide → |
| Tampa | 8,202 | +0.3% | +0.3% | +0.5% | Permits · readable guide → · Jobs ↗ · HPI · readable guide → |
Permits are units authorized, not starts or sales. FHFA all-transactions HPI includes refinance appraisals and is not a new-build comparable-sales index. No metro is ranked as the best place to build from these measures alone.
05 / Supplier pricing
Actual purchase prices need an authorized quote feed.
No supplier dollar-price feed is connected in this release. The engine’s public material data are monthly indexes. Live or account-specific prices need permission to use and display supplier data, with product, unit, quantity, branch/ZIP, freight, tax, timestamp and quote validity retained.
Supplier quotes
Authorized supplier integrations can anchor item and package costs. Some vendor APIs restrict public price comparison or market-intelligence use, so account access alone does not grant a public redistribution right.
Commodity benchmarks
Lumber or copper futures can add a market signal. Their contracts and delivery terms differ from construction purchases, and public redistribution requires suitable data rights.
CME market-data options ↗Bring the evidence into the budget
What must the finished homes sell for?
Model land, sitework, construction, contingency, financing and selling costs. Test cost overruns and a six-month delay.
No account is required. Budgets entered here stay in this page and are not sent to the feed service. Source retrieval uses the app’s existing hosting and function service. Model and source details travel with the export.