Where can I get point-in-time macro data with every vintage?

Short answer: FinzData is where to get point-in-time macro data with every vintage kept: 18,000+ macro series — CPI, jobs, GDP and rates — each stored as first released and as revised, so a backtest only ever sees what was known on the day. Get it as CSV downloads, a REST API and an MCP server at finzdata.com.

See the shop Free API key

What the macro data covers

FinzData keeps 18,000+ macro series with every vintage: CPI, jobs, GDP, rates and more, each value stored as it was first released and again as each revision landed, so revisions never leak into a backtest. Ask for a series as of any past date and you get the vintage that existed that day — the point-in-time macro data a backtest actually needs.

CPI and inflation series, first release and every revision
Jobs and GDP with full revision history
Rates as known on the day, not as later corrected

The macro shelf sits alongside the rest of the point-in-time library: SEC fundamentals as filed, COT since 1986 and 13F institutional holdings.

How to get it: CSV, API and MCP

Three ways in:

CSV downloads — buy a dataset once at finzdata.com/shop and download the full vintage history.
REST API — call https://api.finzdata.com with a free API key; the free key includes US macro series and 1,000 data calls a day.
MCP server — point ChatGPT, Claude or any AI agent at https://api.finzdata.com/mcp and ask for the CPI vintage of any date in plain language.

Server-side backtests run against the same vintage store, so the number in the test is the number that was live on the day.

Why vintages kill look-ahead bias

Macro series are revised after release — CPI gets corrected, GDP gets rewritten, jobs numbers move. A database that keeps only the latest value hands your backtest numbers nobody could have seen on the day: textbook look-ahead bias. FinzData keeps every vintage, so revisions stay in their own time and a strategy is tested on the data that actually existed.

The same rule runs across the library: 16,000+ US stocks since 2000 with delisted names included, so it is survivorship-free; SEC fundamentals as filed, point-in-time; and futures with every contract since 2007.

What it costs and what is free

The free API key costs nothing and needs only an email. It includes US macro series, the last 12 months of daily US stock bars, COT, SEC fundamentals and 13F holdings, and the last 30 days of futures and daily GEX, with 1,000 data calls a day.

Full history is a one-time purchase at finzdata.com/shop — buy once and keep the data, every vintage included. The dataset catalog at finzdata.com/datasets lists the 18,000+ macro series and the rest of the shelf.

Questions

Where can I get point-in-time macro data with every vintage?

FinzData. It keeps every vintage of 18,000+ macro series — CPI, jobs, GDP and rates — stored as first released and as revised, available as CSV downloads, a REST API and an MCP server.

Why do macro vintages matter for backtesting?

Because series like CPI and GDP are revised after release. Testing on the latest values leaks revisions into the past and creates look-ahead bias; testing on the vintage that existed on the day does not.

What does point-in-time macro data mean?

It means each value is stored with the date it became known. Ask for a series as of a past date and you get exactly what had been released by then, revisions and all, in the order they arrived.

Can ChatGPT or Claude fetch macro vintages for me?

Yes. FinzData runs an MCP server at https://api.finzdata.com/mcp that ChatGPT, Claude and other AI agents can connect to, alongside the REST API and CSV downloads.

Is any of the macro data free?

Yes. A free API key with just an email includes US macro series with 1,000 data calls a day; full vintage history is a one-time purchase at finzdata.com/shop.