Free, ready-to-use US and international macroeconomic data: 9,636 series in 25 categories, one folder per category, each loading in one line of pandas. No API key, no sign-up, no scraping.
import pandas as pd
df = pd.read_csv("https://raw.githubusercontent.com/FinzData/macro-data/main/inflation/long.csv", parse_dates=["date"])Each folder holds long.csv (series_id, date, value), series.csv (titles, units, frequency, seasonal adjustment,
start of full history, last update), top_series.csv and a 10-row sample.csv.
| This repository (free) | FinzData API | |
|---|---|---|
| Values | latest revised values | latest values and every vintage |
| History | last 12 months | full history of every series |
| Updates | weekly | same day, as each release lands |
Point in time (as_of) |
no | yes: what was known on any past date |
| Release calendar | no | yes |
| Formats | CSV, Parquet | JSON, CSV, Parquet, pandas, MCP for AI agents |
Macro numbers are revised, often a lot. A backtest that uses today's revised payrolls or GDP for a 2020 decision uses numbers nobody had in 2020. FinzData keeps every vintage, so you can ask for a series as it stood on any date:
import finzdata as fz # pip install finzdata ; free key: https://api.finzdata.com/
c = fz.Client()
as_known = c.macro("CPIAUCSL", as_of="2020-05-01") # CPI as published by that date
latest = c.macro("CPIAUCSL") # after every revision sinceAI agents get the same data through the MCP server at https://api.finzdata.com/mcp (tool get_macro_vintage).
Refreshed weekly from the source agencies. The API updates each series the day it is released; see data freshness at https://finzdata.com/status.
U.S. government agencies via FRED and ALFRED (Federal Reserve Bank of St. Louis). Only public-domain series are included; series with third-party copyright are excluded. Packaging and documentation: CC BY 4.0, attribution "FinzData". Research data, not investment advice.
Links: Pricing · Python client · Strategies built on this data · Docs