Methodology: every metric, defined

1. Metric dictionary

Fourteen metrics account for every number on the site. Each entry gives the formula as we compute or receive it, the provider, the cadence, and one plain sentence on what it actually tells a buyer or seller.

2. Market-condition verdict rules

Every city and ZIP page carries one verdict. It is produced by a deterministic score — the same function, with the same thresholds, for all 22,712 cities and 30,888 ZIP codes. There is no manual override and no per-market tuning.

Start at zero. Add or subtract points for each of the three signals below. Positive points favour sellers, negative points favour buyers. Then map the total to one of five labels.

The five classifications

3. Affordability & payment math

Every payment figure on the site — in the calculators and in the affordability sentences on market pages — comes from the standard amortising-loan formula. Here it is in full.

Monthly principal & interest

When the rate is exactly zero the formula divides by zero, so we fall back to M = P ÷ n. The default rate in every calculator is the most recent Freddie Mac 30-year survey average from FRED, currently 7.03% (week ending September 24, 2026) — you can override it. Freddie Mac via FRED, September 2026 · Refreshes weekly

The full monthly housing cost

Principal and interest is not the payment. We add four components, using these assumptions unless you change them:

The income-needed rule (28/36, or 28/43)

To answer “how much house can I afford” we apply the two ratios most conventional underwriters use, and take the lower of the two answers:

• •Front-end, 28%. Total housing cost may not exceed 28% of gross monthly income. max housing = (annual income ÷ 12) × 0.28
• •Back-end, 36% or 43%. Housing cost plus all other monthly debt payments may not exceed this share of gross monthly income. 36% is the conventional underwriting standard and the affordability calculator’s default; 43% is the Qualified Mortgage ceiling many conforming lenders will stretch to, and is selectable. available = (annual income ÷ 12) × 0.36 − existing monthly debts

The lower of the two caps is the housing budget. Turning that budget into a price is not a subtraction, because property tax and PMI are both percentages of the price we are solving for — they have to be collected on the same side of the equation. Only the insurance premium is a flat dollar amount, so it is the one term that moves to the numerator. Solving exactly:

This replaced an earlier shortcut that reserved a flat 60% of the housing cap for principal and interest (revised 5 September 2026). That approximation over-stated affordability in low-property-tax states and under-stated it in high-tax ones by tens of thousands of dollars; the closed-form solve above has no such bias. Maintenance is deliberately excluded from the qualifying math — no lender escrows it — and is reported as a separate 1%-of-value annual reserve. Every output is an illustration for planning, not a pre-approval: only a lender pulling your credit and verifying your income can tell you what you qualify for.

4. Rent-vs-buy method

The rent-vs-buy comparison simulates both paths month by month over your chosen horizon and then compares them the only way that is like-for-like: net worth if you liquidated on the same date. Rewritten 5 September 2026 — the previous version amortised the loan over the holding period rather than its 30-year term, computed equity against the original loan amount, and ignored the opportunity cost of the down payment entirely. The current steps, in order:

1. 1.Same starting cash: both paths begin with the down payment plus closing costs (3% of price by default). The buyer converts it into a house and fees; the renter invests it. Everything after this point is directly comparable.
2. 2.Cost of renting each month: rent, compounded monthly at the annual rent-growth rate you set (3% by default). Rent is no longer held flat over the horizon — the old model's flat rent was a systematic thumb on the scale for renting.
3. 3.Cost of owning each month: principal & interest on the actual 30-year term (formula above) + property tax at your rate ÷ 12 + insurance ÷ 12 + maintenance at 1%/yr ÷ 12 + PMI at 0.5%/yr while the balance exceeds 80% of the purchase price. Tax and maintenance track the home's current value, not the purchase price.
4. 4.Invest the difference, both directions: whichever path costs less that month invests the difference at the investment-return rate (7% by default). Early on that is normally the renter; once compounding rent passes a fixed mortgage payment it flips to the owner. Both balances compound monthly.
5. 5.Amortise properly: each month the interest portion is the outstanding balance × the monthly rate and the remainder reduces principal. Equity is the balance actually paid down, not an approximation.
6. 6.Liquidate and compare: buy net worth = home value × (1 − cost to sell, 6% by default) − remaining loan balance + the buyer's investments. Rent net worth = the renter's investments. The cost to sell is charged on the full sale price, not on equity.

The answer is whichever net worth is higher at the end of the horizon, and the size of the gap is reported alongside it.

The breakeven month is the first month at which buy net worth reaches or passes rent net worth, both measured as defined in step six. Opportunity cost on the down payment is now modelled explicitly, which is the single largest correction to the old figure — the crossing typically arrives years later than the previous method reported. It remains a nominal comparison: it does not discount future dollars, and it deliberately excludes the mortgage interest deduction, because since the standard deduction was roughly doubled most filers receive no marginal benefit from it and modelling it would systematically favour buying. Capital-gains treatment also differs between the two paths and is not modelled. Treat the breakeven as the centre of a wide distribution, not a date.

5. Source cadence & freshness

Five providers, five different clocks. Every figure on the site is stamped with its provider and period so you never have to guess which clock a number is on.

Full per-provider licensing terms, coverage counts, and links to each provider’s own terms of use are on the data sources page.

6. Known limitations

Six things this data cannot do. We would rather you knew them up front than discovered them after acting on a number.

7. Corrections policy

We publish the rules that produce every number, which means a correction is nearly always traceable to one of three things: a bad row from a provider, a bug in a pipeline, or a rule on this page that is wrong. Here is how each is handled.

• •Provider data error. We do not overwrite a provider's published value. If the provider republishes a corrected figure, our next scheduled ingest picks it up and the page re-renders. If a value is obviously impossible, we suppress the metric on that page rather than display a number we do not believe.
• •Pipeline or rendering bug. Fixed at the source, then the affected pages are re-rendered. Bugs that changed a displayed figure for more than a handful of locations get a dated note in this section.
• •Rule or definition error. Any change to a metric definition, a scoring threshold, or a calculator assumption changes the Last reviewed date at the top of this page, and the previous rule is described in the note so you can tell which version produced a figure you saw earlier.

Found a figure that looks wrong? Send us the URL and the number. Get in touch — we show our work, so we can usually tell you within a day whether the data is wrong or the rule is.