Sources & methodology

Know what you’re looking at.

Useful research starts with clear definitions, visible dates, and an honest account of what is still unknown.

New-build engine

The construction intelligence engine connects monthly national material price indexes and new-home sales/supply with experimental statistical forecasts. Its model card explains selection, independent historical validation, data revisions and error ranges. Material indexes are not supplier quotes. The build-for-sale feasibility model separately uses explicit user assumptions for project costs, financing and selling proceeds; it does not infer local sale prices or peak funding from the public indexes.

Six tools for a build-for-sale decision

The new-build workbench uses a shared feasibility project where indicated. Starting budgets, buyer terms and quotes are illustrative until you replace them with project evidence. Private project inputs stay in browser storage and can be exported. Clearing storage removes saved work; a public source refresh does not submit your budget or supplier documents.

Project stress test: solves selling-price, construction-cost and delay thresholds for the entered target margin, then combines user-defined shocks. Verification priority measures the loss of margin headroom from each entered adverse change. It is a sensitivity comparison, not a probability of failure. Average loan draw remains fixed; this is not a monthly cash-flow schedule.
Buy now, lock or wait: compares manual quotes using whole purchasable packs, freight and an entered effective tax rate, then adds financing, storage, damage and delay assumptions. Future material-price change is a user scenario. A supplier lock is modeled only when its terms are entered. No order is placed or substitution certified.
House cost feed: applies each matching index’s percentage change from the exact baseline month to the uncommitted material budget. Paid and locked costs receive no index adjustment. Missing or unmapped exposure stays visible. The optional project overlay subtracts the covered material change from profit without recalculating financing, contingency or schedule.
Land readiness passport: records your documents, dates, conditions and next actions across six evidence areas, with jurisdiction guides for Raleigh and Charlotte. Statuses reflect your assessment, not a StackX parcel check or agency approval. A user-entered delay can be tested against the linked feasibility project.
Buyer payment to land: solves a home-price ceiling using principal, interest, property tax, insurance, HOA and entered PMI, then connects that price to development costs and a target land allowance. It compares price cuts, quoted permanent-rate buydowns and a temporary 2–1 payment subsidy. Affordability arithmetic does not establish buyer demand or mortgage qualification.
Public forecast record: preserves issued predictions and their original source snapshots for later comparison with observed outcomes. The record’s publication and scoring rules are described below.

Immutable predictions, dated scoring evidence

The public forecast record keeps each original prediction, selected model, source history and historical error range unchanged. It uses the earliest durably stored edition for each UTC calendar month. Opening the archive after a qualifying fresh retrieval can issue that month’s edition; publication is not scheduled daily and is never backdated.

Three- and six-month targets are measured from the last observed source month. A prospective prediction must be issued before its target month begins. Future outcomes remain pending until an exact completed target-month observation is available with matching series, units and geography. Historical backtests are reported separately and do not count as realized prospective accuracy.

Absolute error is the absolute difference between the later observation and the original forecast. The comparison baseline predicts no change from the last value known at issue. Scores use the latest retrieved source vintage, so revisions to the source can revise a score while the issued prediction stays unchanged. Downloads include the original edition and the current scoring evidence. Historical error ranges are not calibrated probability intervals, and unavailable models remain visible.

Two research layers, distinct geographies

The September 2026 housing-cycle report covers six full metropolitan areas using BLS monthly payroll jobs, FHFA quarterly home-price histories and Census building permit authorizations. It includes its own period definitions, formulas, reporting-coverage analysis, original theses and downloadable data. These metro measures are separate from the city-limit figures described below.

City profile dataset

Each city profile uses the U.S. Census Bureau’s QuickFacts publication. Median household income, median gross rent and median owner-occupied home value are American Community Survey 2020–2024 five-year estimates, expressed in 2024 dollars. Population is a separate Census Population Estimates Program estimate for July 1, 2025. These measures must not be treated as observations from the same year.

Figures were checked on September 5, 2026 and stored as a reviewed snapshot. This release does not automatically refresh the city data. Original source links on every profile let you check for later publications. Existing housing-stock values are not new-build sale comparables, and household income alone does not establish qualified buyer demand.

Definitions that matter

Geography: city limits, not the full metropolitan area, county, or a neighborhood. City boundaries and metro market areas differ.
Median household income: annual household income context. It does not identify active buyers, available down payments, credit eligibility or the distribution of buyer budgets.
Median gross rent: a survey estimate of rent including applicable tenant-paid utility costs. It remains available in legacy tools and CSV exports; it does not establish a current asking rent or a property’s potential rent.
Median home value: reported value of owner-occupied housing. It does not establish current transaction prices or comparable closed sales for new construction.
Population: a count estimate describing scale. This release does not calculate population growth or predict housing demand.
Sampling uncertainty: ACS values are estimates. We do not display margins of error in this first release; consult the underlying ACS tables and comparison guidance before drawing conclusions from small differences.

How markets are selected

Charlotte, Raleigh, Dallas, Houston, Phoenix, and Tampa are an initial sample across four states. Inclusion reflects a manageable starting scope for comparison. StackX has not ranked their investment performance or assessed individual properties within them.

Earlier rental tools and educational guides

Earlier articles retain their original educational scope. The rental calculator uses your hypothetical inputs to calculate income after vacancy, operating costs, and principal-and-interest payments. It displays an illustrative downside scenario alongside the base case. It does not forecast appreciation, exit proceeds, tax effects, investment performance, or the probability of any outcome. Review its stated exclusions and replace teaching inputs with property-specific evidence.

The offer-price tool converts a monthly debt-payment budget into a loan principal using a fully amortizing fixed-rate loan. It deducts a replacement reserve separately, while the rental cash-flow calculator shows cash flow before separate reserves. The tools therefore require consistent treatment of reserves when comparing results. At 100% down, cash flow alone cannot determine an offer-price ceiling.

Saved rental scenarios and due-diligence notes stay in browser storage and can be exported as files. Status labels in the notebook record your assessment of evidence; StackX does not verify documents or mark a property suitable to buy. The source library is a directory of official publications and local authorities, each with its own publication and access rules.

What we don’t claim

StackX currently has no listed investment offerings, verified deal pipeline or sponsor track record on this site. Project models do not independently verify a property’s underwriting inputs. A city profile is a research starting point. A saved market is a personal browser bookmark. No part of the interface reserves a property, commits capital, or starts an email subscription.

Go to the source

Spotted an error or a newer publication? Send a correction with the city, indicator, period, and source link.