Historical Macro Economic Data Download
Historical macro economic data for the major currency areas: growth, inflation, labour markets, monetary policy, government bond yields, money and credit.
Every value is kept together with the revisions it went through, so a backtest can use what was known on a given day and not only the final figure. Each series says how far its revision dates can be trusted.
- 228 series
- 897,327 observations with all vintages
- 1913-01-01 → 2026-09-17
- Data updated 2026-09-26
What is in the catalogue
Markets
United States (USD): 51 series; Euro area (EUR): 34 series; Japan (JPY): 25 series; United Kingdom (GBP): 24 series; Australia (AUD): 39 series; Canada (CAD): 29 series; Switzerland (CHF): 9 series; New Zealand (NZD): 9 series; China (CNH): 1 series; Gold (XAU): 2 series; Silver (XAG): 2 series; WTI crude oil (WTI): 2 series; Brent crude oil (BCO): 1 series.
Categories
Additional Indicators (2), Commodity Fundamentals (3), Economy (92), Exchange Rate (11), Government Bond Yields (38), Labor Market (38), Monetary Policy (38), Money & Credit (4), Volatility (2).
Backtesting with point-in-time macro data
Each row in the revision files says which period a value describes (observation_date), the day it became the current figure (realtime_start), the day a newer figure replaced it (realtime_end, 9999-12-31 while it is still the latest) and which revision it is (revision_number 0 is the first print).
To test a strategy as it would have run, join on the release date: on any test date, take for each period the row whose realtime_start is on or before that date and whose realtime_end is after it. Series marked full revisions are safe to use this way. For tracked revisions the first print has no release date, so use them with care before the tracking start.
About the data
- observation_date is the period a value describes. In the revision files, realtime_start is the day a value became the current one and realtime_end is the day a newer value replaced it (9999-12-31 while it is still the latest).
- vintage_basis = source_vintages: the publisher's own vintage history, so realtime_start is the real release date of each value and the file can be used for point-in-time backtests.
- vintage_basis = self_tracked: the first value of each period has an empty realtime_start, because when it was first published is not known. Later revisions are dated by the day we noticed them (from 2026-09-01 for most of these series); revisions before that cannot be recovered.
- vintage_basis = not_revised: one value per period, with both dates empty. No earlier prints exist in our data.
- survey = true marks opinion or survey based series (confidence, PMI). discontinued = true means the source has stopped publishing; the history is kept and marked, never dropped.
Frequently asked questions
What is historical macro data with revisions?
Economic figures such as GDP, inflation or payrolls are published first as an estimate and revised later, sometimes several times. This catalogue keeps every version of every value together with the day it appeared (its vintage), so you can see what the market knew on any date, and not only the final number.
Why do revisions matter for a backtest?
If a test uses the final, revised figure for a past month, it uses information nobody had at the time. That is look-ahead bias, and it makes strategies that trade on data releases look better than they were. Point-in-time data avoids it. Choose the full revision history to test with what was known then, or the latest values only if you just need the current history.
How reliable are the release dates?
It depends on the series and every series says so. 61 series carry the publisher's own vintage history, so every release date is real. 52 series are revised, but we only know the revision dates from the day we started tracking them; the first value of each period has no release date. 115 series have a single value per period. The README inside every download explains this again.
Which economies and indicators are covered?
The catalogue has 228 series for United States, Australia, Euro area, Canada, Japan, United Kingdom, Switzerland, New Zealand, WTI crude oil, Silver, Gold, Brent crude oil and China. They are grouped into economy, government bond yields, labor market, monetary policy, exchange rate, money & credit, commodity fundamentals, additional indicators and volatility. Some series come from surveys (confidence, PMI) and are marked as such, and series that a source has stopped publishing are kept and marked as discontinued.
How far back does the data go?
The earliest observation is from 1913; each series shows its own first and last period in the catalogue above. The most recent data on this page ends 2026-09-17.
What do I get when I download?
A zip file with one CSV per series, a catalogue file describing every series (frequency, unit, how far the revision dates can be trusted) and a README. Choose either every vintage of every value, or the latest value of each period.
How often is the data updated?
Once a week. New releases and revisions are added, and the whole set is compared with its source again before it replaces the previous copy.
More data
- Forex Candlesticks – One-minute candles for FX pairs, metals, oil and stock indices, with M5 to monthly bars built from them. Ready for MetaTrader 4 and 5.
- COT Reports – Weekly Commitments of Traders reports for currencies, metals, energy and agricultural futures: how commercials, funds and dealers are positioned, back to 1986.
- All data – the Database overview.
