Numbers need
their context.

Published benchmarks, defined populations.

Mortgage surveys. The site displays Freddie Mac PMMS 30-year and 15-year weekly national mortgage rate averages from Freddie Mac's publisher-linked current-data file under its published automatic-data website-use terms. The figures are unchanged and source-dated, with Freddie Mac attribution; they are not live lender quotes or personal offers. No PMMS history, derived changes, calculator prefill, email or AI use is enabled. Freddie Mac PMMS.

Rates & policy. Treasury’s official annual XML package supplies daily par yields at 3 months and 2, 5, 7, 10 and 30 years. Curve spreads subtract same-date short yields from the 10-year yield. These are not MBS spreads. EFFR uses the New York Fed’s direct Markets API; target bounds and bank-credit readings use the Board’s official HTML tables. Each direct delivery has its own reviewed reuse terms, not blanket FRED permission.

Agency data. The desk uses the BLS keyless API, BEA’s current published release tables, Labor Department national weekly claims XML, and FHFA’s monthly master CSV. Only national, traditional purchase-only FHFA observations are selected. Agency data are attributed; logos and third-party content are not reused.

Observation date is not fetch time.

The observation date identifies the period the value represents. For a weekly survey, its publication date may be shown separately when the source makes it explicit. For daily indicators, the observation date is the dated market record.

“Last checked” describes when Studio 1003 last retrieved data. A provider's metadata update time is a separate timestamp and does not imply that the underlying observation was republished then.

Changes are expressed in basis points against the previous available observation. One basis point is one hundredth of a percentage point. A weekly survey compares with its prior survey; a daily series compares with its previous business observation.

History only when it is real and permitted.

Historical views use dated source observations without filling weekends, holidays, or gaps. Points are placed by their actual dates. We do not smooth a line or invent daily survey values between weekly releases.

Available time ranges are limited to periods supported by real observations. High and low values describe only the selected, displayed observation window. A table provides the same records in an accessible format.

Levels, growth and estimate vintages.

CPI monthly growth uses seasonally adjusted indexes; CPI 12-month growth uses unadjusted indexes. PCE monthly and 12-month price changes are explicitly reported by BEA. A prior PCE 12-month comparison is withheld when absent from the current release summary. PPI core means final demand less foods, energy and trade services—not merely less food and energy.

Payrolls are a monthly level difference in thousands, JOLTS is a level in thousands, earnings are all-employees total-private 12-month growth, and ECI is civilian total-compensation 3-month index growth. Claims show each series’ latest reported week; continuing claims lag initial claims. Blank future rows in the annual claims export are not releases. FHFA growth uses the national seasonally adjusted purchase-only index, not all transactions.

Growth equals (current index ÷ comparison index − 1) × 100. Monthly and quarterly calculations require contiguous periods; 12-month calculations require all intervening months. No interpolation, forward filling or future observations. Prior readings use the same retrieved revision vintage. A missing latest input never silently becomes an older current reading. Daily yields compare the previous available business observation. Yield changes and curve spreads are basis points; differences between percentage readings are percentage points. GDP is a seasonally adjusted annual growth rate with advance, second or third estimate identified; it is not unannualized quarterly growth. Prior GDP is used only when explicitly stated in the same release.

Publication time and retrieval time are separate. BLS timestamps are shown only for separately verified matching releases. Treasury, claims and FHFA downloads do not establish a publication time. Current means latest retrieved vintage, not a real-time or complete-market claim. Source footnotes retain preliminary and revision caveats; a missing preliminary flag is not a claim of finality.

Calendar rows are explicit official BLS, BEA, Census, FHFA and Federal Reserve schedules, manually reviewed October 17, 2026 for added families. Older retained CPI/jobs/FOMC rows show their own review dates. BEA announcements can supply a next release. Re-reviewed event identities replace older versions. Dates are never generated from recurring weekdays. FHFA dates have no verified time and display as dates only; all supplied times use Eastern daylight-saving rules. Browse the full upcoming list by scrolling. Future PPI/JOLTS, weekly claims and bank-credit schedules remain unverified here; absence is not a claim of no release.

Requests use single-flight, process-local caches, 8-second HTTP timeouts, bounded decoded bodies, fixed agency hosts and no redirects. BLS sends 11 series in one request, refreshing normally every 6 hours, at most 20 attempts per rolling day (including failures). BEA PCE and GDP each use discovery plus release requests; PCE refreshes at most every 6 hours normally and every 15 minutes when due; GDP has a 12-request/day budget. Treasury refreshes hourly (30 attempts/day), claims every 6 hours (12/day), FHFA daily (4/day). At most two additional providers run concurrently alongside BLS and PCE. Failures are cached for 15 minutes. Last-good BLS/PCE/Treasury outputs expire after 24 hours; claims/FHFA/GDP after seven days and are labeled stale on refresh failure. Observation-age safety limits also apply. These quotas are not shared across processes or restarts. No new database or distributed-cache guarantee is claimed.

Sources and reuse: BLS API · BLS public domain · BEA public domain · DOL reuse · FHFA reuse · FRED service terms. No new indicators are sent to generative AI, newsletter or blog ingestion.

Census housing and retail definitions.

The authorized free Census Economic Indicators API supplies eight readings in Development. Production ingestion remains disabled pending separate approval. Starts and permits select total and single-family units; new-home sales select single-family houses sold. These five levels are in thousands at seasonally adjusted annual rates. New-home inventory is thousands of houses for sale, seasonally adjusted but not annualized. Months’ supply is a seasonally adjusted number of months, not an annual rate.

Retail and food services growth is calculated from advance monthly sales in millions of current dollars, seasonally adjusted. It is nominal, not inflation-adjusted, and not annualized. Negative growth is retained. Current and prior growth use adjacent months from the same retrieved vintage; missing inputs are never filled. The API does not establish a publication time or identify individual preliminary/revision flags; estimates can be revised.

Three program packages each request six completed months, refresh every six hours and allow at most eight attempts per rolling day per process, including failures. Each response is bounded to 256,000 decoded bytes with an eight-second timeout and no redirects or automatic retries. Failures are cached for 15 minutes; retained last-good readings are labeled stale and expire after seven days. Missing keys, invalid responses and program outages remain explicit and do not replace other providers’ readings. Census limits remain process-local, not a guarantee across instances or restarts; Census has not been added to shared storage.

This product uses the Census Bureau Data API but is not endorsed or certified by the Census Bureau. See the Census API terms and free key guidance. No Census inputs are sent to AI, newsletters or blog generation.

Fed policy and bank-credit sources.

Direct access and reuse terms were reviewed October 17, 2026. Previously blocked Board HTML tables now respond through ordinary HTTPS requests; no restricted downloads are bypassed and no FRED series is substituted. If delivery, validation or required rights fail, the reading stays unavailable, or a retained reading is explicitly stale.

Effective Federal Funds Rate (EFFR). The New York Fed calculates the volume-weighted median of overnight federal funds transactions reported in FR 2420. The direct API’s effective date is the transaction date; publication normally follows the next business day. It is not a target bound, SOFR or a monthly average. Daily changes are Studio 1003 percentage-point calculations.

The Effective Federal Funds Rate (EFFR) is subject to the Terms of Use posted at newyorkfed.org. The New York Fed is not responsible for publication of the EFFR by Studio 1003, does not sanction or endorse any particular republication, and has no liability for your use. Studio 1003 is not affiliated with the New York Fed. The New York Fed does not sanction, endorse, or recommend any products or services offered by Studio 1003.

© 2026 Federal Reserve Bank of New York. Content from the New York Fed subject to the Terms of Use at newyorkfed.org. Source available without charge. The terms permit business distribution, storage and derived works, with source notices, no endorsement and the same permissions, conditions and restrictions passed downstream. Studio 1003 labels its calculated changes separately and imposes no additional restrictions on the source content.

FOMC target bounds. The Board’s open-market policy history supplies both bounds. Dates are effective dates, not announcement dates. The previous comparison is the preceding policy setting, not the previous day. Future-effective settings are excluded. A setting does not expire merely because it is more than seven days old; verification is bounded by cache retention.

Delinquency. The Board’s All Banks, SA table, domestic-office residential column, describes loans secured by one- to four-family properties, including home equity lines of credit, at insured U.S.-chartered commercial banks. Delinquent means 30 days or more past due or in nonaccrual status. This is not all mortgage lenders, multifamily loans or only smaller banks.

SLOOS. The latest release’s chart-data Figure 3, Panel 1 GSE-eligible column supplies the net percentage of domestic respondents tightening residential mortgage standards: tightening minus easing. Negative values mean net easing. This is not loan demand, subprime, or an other-bank-only subset. The quarterly label is the source survey quarter, which can be published before quarter end, not the preceding quarter of lending activity. Three-or-fewer-respondent suppression stays missing.

Board-origin numerical policy, delinquency and SLOOS tables are public domain under the Board’s copyright policy, unless otherwise indicated. Copying, public-commercial display, caching and calculated changes use those numerical data with Board attribution, not logos or third-party material. Changes are Studio 1003 calculations; no agency endorsement is implied. All five readings retain current/prior observations from the same retrieved vintage; quarterly comparisons require adjacent quarters and no missing values are filled. These downloads do not establish a publication timestamp.

EFFR and policy refresh hourly, at most 30 attempts each per rolling day/process, retaining last-good output for 24 hours. Delinquency and SLOOS refresh daily, at most 4 and 8 HTTP attempts respectively; SLOOS uses index discovery plus one chart table. Their last-good retention is seven days. All use the existing two-provider concurrency limit, eight-second HTTP timeout, two-megabyte decoded-body bound, single-flight caches and 15-minute failure caching. Denied rights suppress cached data immediately. These limits are process-local, not shared across restarts. FRED caching/AI terms are not used as permission for direct PMMS website display; the FRED PMMS gates and SOFR gate remain unchanged.

Finite coverage checklist.

Every selected target is listed below. “Connected” identifies a reviewed delivery path, not a guarantee the source is currently available; the market desk shows each retrieval’s actual status. Missing access, failed requests and invalid source data remain explicit limitations, not substitute live widgets.

Not covered: live MBS/TBA/UMBS, current coupons, OAS, dollar rolls, lender rate sheets, rate forecasts or lock recommendations. MBA, NAR, Case-Shiller, ISM and consumer-confidence data are not licensed here. The desk does not claim coverage of all mortgage-industry data.

Permission shapes the display.

Source attribution does not itself grant redistribution rights. Studio 1003 applies source-specific display and history permissions before publishing a value or historical series. If a record is unavailable, restricted, stale, or otherwise not cleared for display, the board identifies its availability without substituting an estimate.

Public-domain Treasury data is attributed to its source. Survey and other provider content is displayed only where the applicable review allows it. Internal validation data is not automatically a public permission.

These figures are for market context and education. They are not an offer to lend, a rate lock, or individualized financial advice.