FRED

What it is

FRED is the Federal Reserve Bank of St. Louis's economic database, and the default place a macro series comes from when nobody is paying for a terminal. On 14 September 2026 its own tag listing counted 851,100 series from 126 sources — the BLS, the BEA, the Census Bureau, the EIA, the OECD, the BIS, a dozen central banks and a long tail of named academics.

The API lives at api.stlouisfed.org/fred/ and is organised the way the database is: series, series/observations and series/search for the data, category and tags for browsing, releases and releases/dates for the publication calendar, sources for provenance, and a Maps set for regional shapes. A second version adds one endpoint, fred/v2/release/observations, for pulling every series on a release in bulk with its full history.

Point-in-time data is not a separate service. The API overview says users query the FRED and ALFRED databases through the same interface, and vintages come back from the ordinary observations endpoint via realtime_start, realtime_end, vintage_dates and output_type. ALFRED is a second front end on one database and one key — where you go to backtest against what was known on a date, without signing up twice.

Pricing

Free, with no pricing page, because nothing here is for sale. A key goes to anyone who opens an account, and the registration page lists API access among the reasons to open one without quoting a fee.

The number to plan against is the throttle. The error reference states that up to 120 requests per minute are allowed before a 429, and that not complying can result in a temporary block. That figure appears nowhere in the Terms of Use, which reserve the right to impose or adjust limits at any time — so treat 120 as the current setting, not a contract.

Data & coverage

Eight top-level categories: Money, Banking & Finance; Population, Employment & Labor Markets; National Accounts; Production & Business Activity; Prices; International Data; U.S. Regional Data; Academic Data.

In market terms that is rates, currencies, commodities and index levels rather than instruments: Treasury constant-maturity yields from the Fed's own H.15 (DGS10), spot exchange rates from H.10 (DEXUSEU), WTI crude from the EIA (DCOILWTICO), the S&P Cotality Case-Shiller national home price index (CSUSHPINSA) and the S&P 500 daily close (SP500). No single-name equities, no ETFs, no options.

Two things beat a file download: units, frequency and aggregation_method transform and re-aggregate a series server-side, and one request returns up to 100,000 observations as JSON, XML, CSV or Excel — most series in a single call.

Integrations

A free first-party Excel add-in, shipped through the Microsoft store for Excel 2019 or later and Microsoft 365 on both Windows and Mac, which refreshes a sheet against new releases. Official iOS and Android apps. Beyond that the vendor ships nothing: no client library in any language, no webhooks, no MCP server. The packages everyone uses are third-party — fredapi on PyPI by Mortada Mehyar, fredr on CRAN by Sam Boysel — and the FRED MCP servers in circulation are independent projects that say so themselves.

Limitations

• A macro series is not a tradable price. SP500 is one daily close, with no intraday bar, no bid or ask and no volume, and the agreement with S&P Dow Jones Indices carries only ten years of it. Nothing here is executable.
• The release lag is measured in weeks. Real GDP for the quarter that ended 30 June 2026 first appeared on 26 August; Case-Shiller's June reading was published on 27 August. Even the daily FX series runs behind — on 14 September 2026 DEXUSEU stopped at 4 September.
• Revisions rewrite history in place. That same GDP quarter had its next release set for 30 September 2026: the number pulled in August is not the number the series holds in October. A backtest reading today's values is looking at figures nobody had at the time — use the vintage parameters, or accept that the result is fiction.
• This is an aggregation, not a primary source. The Fed republishes other agencies' numbers under its own series IDs; the definition, the seasonal adjustment and the errata belong to whoever published them. When a figure looks wrong, the fix is at the BLS or the BEA.
• Free container, copyrighted contents. The legal notice says in capitals that you must contact the data owner before using third-party series for anything beyond personal use, and the S&P 500 notes prohibit reproduction in any form without prior written permission. Copyrighted series are identifiable: the word Copyright appears in their notes.
• The terms constrain what you build. You may not replicate the essential user experience of FRED, and any application using the API must prominently display the notice that it uses the FRED API but is not endorsed or certified by the Federal Reserve Bank of St. Louis.

Alternatives

For macro alone there is little reason to pay. Nasdaq Data Link once hosted much of this material and has since narrowed to paid products, which is the argument for coming here instead. OpenBB wraps FRED rather than replacing it, alongside the SEC, ECB, IMF and OECD. Koyfin is for when you want the series next to equities and fundamentals in a research product rather than a notebook.