forex-centuries

The most comprehensive open-source collection of long-run financial and economic data available. 27 sources, 1,100+ files, ~240 countries, spanning twenty centuries (1 CE-2026). Exchange rates, gold, silver, interest rates, commodity prices, CPI, GDP per capita, real wages, sovereign debt, regime classifications, and real effective exchange rates — with an automated build pipeline, weekly CI updates, and reproducible analysis.

No other free repository combines this breadth of asset classes across this depth of history. The only comparable product is Global Financial Data (commercial, institutional pricing). The closest open alternative, the Global Macro Database (111 sources, 46 variables), covers macroeconomic aggregates but has no gold, silver, commodity prices, interest rate series spanning centuries, medieval exchange rates, daily FX, or crisis indices.

Related project: fatcrash — crash detection via fat-tail statistics (LPPLS, EVT, Hill estimator, Kappa).

Data overview

Category	What	Countries/Series	Period
GDP per capita	Real GDP per person	178 countries	1 CE - 2022
Exchange rates	Bilateral FX vs USD, GBP, and cross rates	~243 countries	1106 - 2026
Gold	Gold prices in GBP and USD	Global	1257 - 2025
Silver	Silver prices and gold/silver ratio	Global	1687 - 2025
Commodity prices	973 historical + 70 modern series (wheat, rye, oil, metals, agriculture)	European/Asian cities + global	1260 - present
Interest rates	Real and nominal, short and long term	8 countries (historical), UK+US (nominal)	1311 - 2025
Inflation / CPI	Consumer price indices	~180 countries	1500 - 2025
Real wages	Labourers' real wage in subsistence ratios	~40 countries	1820 - 2000+
Sovereign debt	Public debt-to-GDP ratios	191 countries	1800 - 2016
FX regimes & crises	Peg/float/fall classifications, banking/currency/debt crisis indices	~190 countries	1800 - 2021
Real effective exchange rates	Trade-weighted, CPI-deflated REER	178 countries	1960s - 2026
Swedish historical macro	FX, CPI, wages, GDP, money supply, stocks, bonds	Sweden	1277 - 2026
Historical financial statistics	Official/market FX, interest rates, money supply, central bank balance sheets	Various	~1500 - 1950
Macroeconomic panel	59 variables (FX, CPI, rates, GDP, credit, housing, stocks)	18 advanced economies	1870 - 2017

Time depth: 2,000+ years for GDP (Maddison, 1 CE), 900 years for medieval exchange rates (MEMDB, 1106), 768 years for gold (MeasuringWorth, 1257), 707 years for real interest rates (Schmelzing, 1311), 654 years for commodity prices (Allen-Unger, 1260), 526 years for exchange rate panels (Clio Infra, 1500), 55 years for daily FX (FRED, 1971).

Citation

If you use this dataset in your research, please cite:

@misc{forex-centuries,
  author       = {Federico Carrone},
  title        = {forex-centuries: Twenty Centuries of Exchange Rate, Gold, Silver, Interest Rate, and Commodity Price Data (1 CE--2026)},
  year         = {2025},
  publisher    = {GitHub},
  url          = {https://github.com/unbalancedparentheses/forex-centuries}
}

Federico Carrone, forex-centuries: Twenty Centuries of Exchange Rate, Gold, Silver, Interest Rate, and Commodity Price Data (1 CE-2026), 2025. https://github.com/unbalancedparentheses/forex-centuries

Please also cite the underlying sources relevant to your work (see Sources below).

Coverage

Asset                 0    500   1100   1200   1300   1400   1500   1600   1700   1800   1900   2000
                      :      :      :      :      :      :      :      :      :      :      :      :
EXCHANGE RATES        :      :      :      :      :      :      :      :      :      :      :      :
Medieval FX           :      :      ██████████████████████████████████████████████████████████      :
Clio Infra FX         :      :      :      :      :      :      ████████████████████████████████████████
MeasuringWorth FX     :      :      :      :      :      :      :      :      :      █████████████████
BoE Millennium        :      :      :      :      :      :      :      :      :      ██████████████████
JST Macrohistory      :      :      :      :      :      :      :      :      :      :     ████████████
Riksbank              :      :      :      :      :      :      :      :      :      :       ██████████
Bruegel REER          :      :      :      :      :      :      :      :      :      :          ███████
Penn World Table      :      :      :      :      :      :      :      :      :      :         ████████
BIS                   :      :      :      :      :      :      :      :      :      :            █████
IMF / WB / GMD / FRED :      :      :      :      :      :      :      :      :      :            █████
                      :      :      :      :      :      :      :      :      :      :      :      :
PRECIOUS METALS       :      :      :      :      :      :      :      :      :      :      :      :
Gold (GBP)            :      :      :      █████████████████████████████████████████████████████████████
Gold (USD)            :      :      :      :      :      :      :      :      :      █████████████████
Silver                :      :      :      :      :      :      :      :      ████████████████████████
LBMA Gold+Silver      :      :      :      :      :      :      :      :      :      :      :  ████████
Gold/Silver Ratio     :      :      :      :      :      :      :      :      ████████████████████████
                      :      :      :      :      :      :      :      :      :      :      :      :
COMMODITIES           :      :      :      :      :      :      :      :      :      :      :      :
Allen-Unger (973)     :      :      :      ██████████████████████████████████████████████      :      :
WB Pink Sheet (~70)   :      :      :      :      :      :      :      :      :      :          ███████
                      :      :      :      :      :      :      :      :      :      :      :      :
INTEREST RATES        :      :      :      :      :      :      :      :      :      :      :      :
Schmelzing (8 ctry)   :      :      :      ████████████████████████████████████████████████████████████
UK+US Nominal Rates   :      :      :      :      :      :      :      :      ████████████████████████
                      :      :      :      :      :      :      :      :      :      :      :      :
INFLATION / CPI       :      :      :      :      :      :      :      :      :      :      :      :
US CPI                :      :      :      :      :      :      :      :      :      █████████████████
Clio Infra CPI        :      :      :      :      :      :      ████████████████████████████████████████
                      :      :      :      :      :      :      :      :      :      :      :      :
GDP PER CAPITA        :      :      :      :      :      :      :      :      :      :      :      :
Maddison (178 ctry)   █████████████████████████████████████████████████████████████████████████████████
                      :      :      :      :      :      :      :      :      :      :      :      :
REAL WAGES            :      :      :      :      :      :      :      :      :      :      :      :
Clio Infra Wages      :      :      :      :      :      :      :      :      :      ██████████████████
                      :      :      :      :      :      :      :      :      :      :      :      :
SOVEREIGN DEBT        :      :      :      :      :      :      :      :      :      :      :      :
IMF HPDD (191 ctry)   :      :      :      :      :      :      :      :      :      ██████████████████
Reinhart-Rogoff       :      :      :      :      :      :      :      :      :      ██████████████████
                      :      :      :      :      :      :      :      :      :      :      :      :
REGIMES / CRISES      :      :      :      :      :      :      :      :      :      :      :      :
IRR Classifications   :      :      :      :      :      :      :      :      :      :       █████████
RR Crisis Indices     :      :      :      :      :      :      :      :      :      ██████████████████
                      :      :      :      :      :      :      :      :      :      :      :      :
SWEDEN (1277+)        :      :      :      :      :      :      :      :      :      :      :      :
Riksbank Hist FX      :      :      :      :      :      ██████████████████████████████████████████████
Riksbank Hist CPI     :      :      :      :      :      :      :      :      :      ██████████████████
Riksbank Hist Wages   :      :      :      :      :      :      :      :      :      ██████████████████
                      :      :      :      :      :      :      :      :      :      :      :      :
CFS HIST. FIN. STATS  :      :      :      :      :      :      ████████████████████████████      :

What the data shows

Fat tails are universal and persistent

Every single currency pair — across all time scales and all centuries — shows heavier tails than a Gaussian distribution.

Daily (1971-2025, 23 currencies): 3-sigma events happen 3-6x more often than a normal distribution predicts. Even the most "stable" pair (EUR/USD, excess kurtosis 2.5) has 4x too many tail events. Pegged and managed currencies (HKD, CNY, LKR) show the highest kurtosis — the peg suppresses daily moves but creates catastrophic jumps when it breaks.

Yearly (1791-2025, 40 currencies): Germany's Weimar hyperinflation produces kurtosis of 36.8 (a single year, 1923, saw a 16x log move). Latin American currencies (Mexico 82, Argentina 19, Brazil 13) show the fattest tails from repeated devaluations and redenominations. Even the UK, with 234 years of data, has excess kurtosis of 5.1.

Currency	Ann Vol	Excess Kurtosis	Tail Ratio
GBP	9.4%	6.9	4.8x
JPY	10.1%	9.0	5.4x
CHF	11.0%	14.5	4.5x
EUR	9.2%	2.5	4.0x
BRL	15.6%	13.1	5.5x
KRW	10.8%	139.7	4.4x

The peg paradox

Currencies with the lowest daily volatility (HKD at 3.2%, CNY at 8.2%) have some of the highest excess kurtosis (261 and 3846). Pegs compress the distribution most of the time but produce massive outliers when they break. This is the classic problem with using volatility as a risk measure — it underestimates the probability of extreme moves in managed currencies.

Volatility clusters by regime

Three distinct regimes emerge from the data:

Gold standard era (~1870-1914): low nominal volatility, sudden large breaks
Bretton Woods (1944-1971): artificially suppressed vol, then explosive devaluations
Free float (1971-present): higher day-to-day vol but fewer catastrophic jumps

The regime-conditional statistics quantify this: freely falling currencies have annual volatility of 225% vs 10.8% for free-floating. Pegged currencies show excess kurtosis of 133 vs 0.8 for free float — confirming that pegs suppress daily volatility but produce catastrophic jumps.

Cross-currency correlations

Daily log-return correlations reveal geographic clustering: Scandinavian currencies (DKK, SEK, NOK) move together, as do Asian managed currencies (SGD, TWD, THB). European currencies are tightly correlated with each other but less with emerging market pairs.

Implications

The data strongly supports modeling FX returns with fat-tailed distributions (stable, Student-t, or power-law) rather than Gaussian. Standard VaR and options pricing models systematically underestimate tail risk in currency markets.

Quickstart

# 1. Install dependencies
nix develop                      # Nix flake (recommended)
pip install -r requirements.txt  # pip fallback

# 2. Explore the data
python quickstart.py             # pure stdlib, no dependencies
python quickstart_pandas.py      # pandas version

Yearly panel: 24,656 obs, 243 countries, 1500-2025
Sources: MW=3,444 | CI=9,031 | GMD=12,181

Longest series:
  United States              526 years (1500-2025)
  United Kingdom             236 years (1789-2025)
  Denmark                    235 years (1791-2025)

Daily: 13,802 dates x 23 currencies (1971-2025)
Medieval: 13,197 Spufford (521 places) + 50,559 Metz records (29 places)

Build pipeline

All derived data and charts are reproducible from source files:

python build.py               # regenerate data/derived/ from data/sources/
python validate.py            # run data quality checks (52 checks)
python visualize.py           # generate charts/ (9 PNGs)

make all                      # build + validate + visualize in one step
make update-sources           # fetch all remote data sources

The 8-step build pipeline produces:

FRED daily normalization (23 currencies, foreign-per-USD convention)
Yearly unified panel (243 countries, MW > CI > GMD priority merge)
Log returns (daily and yearly)
Volatility statistics (kurtosis, tail events, 3-sigma counts)
Correlation matrices (daily 23x23, yearly 40x40)
Rolling volatility (252-day window)
Regime analysis (IRR fine->coarse, regime-conditional stats)
Gold inflation (yearly 243 countries since 1257, monthly 174 currencies)

Testing

pytest tests/ -v              # 13 unit tests for build pipeline correctness
make test                     # same via make

Tests cover FRED inversion logic, source priority merge, log return formulas, tail event counting, regime mapping, gold calculations, and rolling window behavior — all using synthetic data with no network calls.

Interactive notebook

jupyter lab notebooks/exploration.ipynb

Seven sections: yearly panel, daily data, fat tails (histogram + QQ-plot), regime analysis, gold inflation, and medieval data. Each section loads the relevant dataset and produces inline charts.

Updating source data

23 sources are automatically fetched by update_sources.py. A weekly GitHub Actions workflow runs --all every Monday at 06:00 UTC, or trigger it manually from the Actions tab. Each run creates a GitHub Release with a tarball of all data.

python scripts/update_sources.py --all           # update all 23 sources
python scripts/update_sources.py --fred          # or update individually

FRED requires a free API key from fred.stlouisfed.org. Set it as FRED_API_KEY in your environment or as a GitHub repository secret. All other sources require no authentication.

Charts

python visualize.py generates 9 PNGs in charts/:

Chart	Description
`fat_tails_histogram.png`	EUR/USD daily log returns vs fitted normal — shows excess density in the tails
`peg_paradox.png`	Annualized volatility vs excess kurtosis — low-vol pegs cluster at extreme kurtosis
`tail_ratio_bar.png`	Observed 3-sigma events / Gaussian expectation — every currency exceeds 1.0
`rolling_volatility.png`	1-year rolling vol for GBP, JPY, CHF, EUR with Nixon shock, Lehman, COVID markers
`correlation_heatmap.png`	Hierarchically clustered daily log-return correlations (23x23)
`gold_erosion.png`	Cumulative gold purchasing power retained for USD, GBP, JPY, CHF, FRF, DEM, INR, CNY
`regime_timeline.png`	Exchange rate regime (peg -> free float -> freely falling) for 25 countries, 1940-2019
`qq_daily.png`	QQ-plots for 6 daily currencies — tail deviation from the normal reference line
`qq_regimes.png`	QQ-plots comparing peg vs free float vs freely falling return distributions

Project structure

forex-centuries/
├── data/
│   ├── sources/           # Raw data, untouched from provider (27 sources)
│   │   ├── maddison/      # GDP per capita, 178 countries (1 CE - 2022)
│   │   ├── memdb/         # Medieval exchange rates (1106-1800)
│   │   ├── allenunger/    # 973 commodity price series (1260-1914)
│   │   ├── measuringworth/# FX, gold, interest rates, CPI (1257-2025)
│   │   ├── riksbank_hist/ # Swedish FX, CPI, wages, GDP, money supply (1277-2020)
│   │   ├── schmelzing/    # Real interest rates, 8 countries (1311-2018)
│   │   ├── cfs/           # Historical Financial Statistics (FX, rates, money, ~1500-1950)
│   │   ├── clio_infra/    # FX, inflation, bonds, debt, GDP, wages (1500-2016)
│   │   ├── freegold/      # 768-year gold, silver, gold/silver ratio (1258-2025)
│   │   ├── lbma/          # Daily gold + silver prices in USD/GBP/EUR (1968-2025)
│   │   ├── reinhart_rogoff/ # Debt/GDP, inflation, crises, gold standard (1800-2016)
│   │   ├── imf_hpdd/      # Sovereign debt-to-GDP, 191 countries (1800-2015)
│   │   ├── bruegel/       # Real effective exchange rates, 178 countries (1960s-2026)
│   │   ├── imf/           # 173 currencies, monthly (1955-2025)
│   │   ├── bis/           # Bilateral + effective rates, ~190 economies (1957-2026)
│   │   ├── fred/          # 23 daily pairs + 2 USD indices (1971-2025)
│   │   ├── gold/          # Monthly gold prices USD (1833-2025)
│   │   ├── riksbank/      # 53 SEK bilateral series (1900-2026)
│   │   ├── worldbank/     # Official rates, all members (1960-present)
│   │   ├── worldbank_commodities/ # ~70 commodity prices, monthly + annual (1960-present)
│   │   ├── pwt/           # Penn World Table: 185 countries, FX + PPP (1950-2023)
│   │   ├── irr/           # Exchange rate regime classifications (1940-2021)
│   │   ├── jst/           # Macrohistory: 18 countries, 59 variables (1870-2017)
│   │   ├── boe/           # UK millennium dataset (1791-2016)
│   │   └── gmd/           # 243 countries, USDfx + REER (1960-2024)
│   └── derived/           # Computed by build.py (never hand-edited)
│       ├── normalized/    # Unified panels (yearly 243 countries, daily 23 currencies)
│       └── analysis/      # Log returns, volatility, correlations, regimes, gold
├── charts/                # Generated by visualize.py (9 PNGs)
├── notebooks/             # Jupyter exploration notebook
├── scripts/               # update_sources.py (23 automated fetchers)
├── tests/                 # Unit tests (13 tests, synthetic data)
├── build.py               # 8-step ETL pipeline
├── validate.py            # Data quality checks (52 checks)
├── visualize.py           # Chart generation (9 PNGs)
├── quickstart.py          # Explore data with pure stdlib
└── quickstart_pandas.py   # Explore data with pandas

Sources

Automated (`update_sources.py --all`)

Source	Flag	Data	Period	Auth
FRED	`--fred`	23 daily FX pairs + 2 USD indices	1971-2025	`FRED_API_KEY`
IMF IFS	`--imf`	173 currencies vs USD, monthly	1955-2025
BIS	`--bis`	Bilateral + effective rates, ~190 economies	1957-2026
Sveriges Riksbank	`--riksbank`	53 SEK bilateral series, daily	1900-2026
World Bank FX	`--worldbank`	Official rates, all member countries	1960-present
World Bank Commodities	`--commodities`	~70 commodity prices (oil, metals, agriculture), monthly + annual	1960-present
JST Macrohistory	`--jst`	18 economies, 59 macro/financial variables	1870-2017
Penn World Table	`--pwt`	185 countries, exchange rates + PPP	1950-2023
MeasuringWorth	`--measuringworth`	41 FX vs USD, gold (5 series), interest rates (UK+US), US CPI, $/£	1257-2025
Clio Infra	`--clio`	FX (USD+GBP), inflation, bonds, debt, GDP, gold standard, real wages	1500-2016
FreeGoldAPI	`--freegold`	768-year gold prices, gold/silver ratio, silver prices	1258-2025
LBMA	`--lbma`	Daily gold + silver in USD, GBP, EUR	1968-2025
DataHub Gold	`--gold`	Monthly gold prices USD	1833-2025
IRR	`--irr`	De facto FX regime classifications, ~190 countries	1940-2021
Bank of England	`--boe`	UK millennium dataset ($/£, rates, prices, GDP)	1791-2016
Schmelzing (BoE)	`--schmelzing`	Real interest rates, 8 countries (IT, UK, NL, DE, FR, ES, JP, US)	1311-2018
Maddison Project	`--maddison`	GDP per capita, 178 countries (via OWID API + Dataverse xlsx)	1 CE-2022
Allen-Unger GCPD	`--allenunger`	973 commodity price series (wheat, rye, silver, etc.) across cities	1260-1914
Bruegel/Darvas REER	`--bruegel`	Real effective exchange rates, 178 countries, monthly	1960s-2026
IMF HPDD	`--imfhpdd`	Sovereign debt-to-GDP, 191 countries (via DBnomics)	1800-2015
CFS HFS	`--cfs`	Exchange rates, interest rates, money supply, central bank balance sheets	~1500-1950
Riksbank Hist. Monetary	`--riksbank_hist`	Swedish FX, CPI, wages, GDP, money supply, stocks/bonds (Vols I-III)	1277-2020
Reinhart-Rogoff	`--reinhartrogoff`	Regime classifications, debt/GDP, inflation, crisis indices, gold standard dates	1800-2016

Static (committed to repo, updated manually)

Source	Data	Period	Notes
MEMDB Spufford	Medieval exchange quotations (521 places)	1106-1500	No export API
MEMDB Metz	Lower Rhine + European comparison (29 places)	1350-1800	No export API
Global Macro Database	243 countries, USDfx + REER	1960-2024	Email-gated

Longest series by asset type

Asset	Series	Years	Period	Source
GDP per capita	178 countries	2,021	1 CE-2022	Maddison Project
Gold	British official price (GBP/oz)	768	1257-2025	MeasuringWorth
Gold	USD price per oz	767	1258-2025	FreeGoldAPI
Interest rates	Real rates, 8 countries	707	1311-2018	Schmelzing (BoE)
Medieval FX	European exchange quotations	694	1106-1800	MEMDB
Commodity prices	973 series across European/Asian cities	654	1260-1914	Allen-Unger
Exchange rates	186 countries vs GBP	516	1500-2016	Clio Infra
Silver	Silver-normalized prices	337	1688-2025	FreeGoldAPI
Gold/silver ratio	Annual ratio	338	1687-2025	MeasuringWorth
Interest rates	UK+US nominal short + long term	296	1729-2025	MeasuringWorth
US CPI	Consumer price index	251	1774-2025	MeasuringWorth
41 FX vs USD	Annual rates	234	1791-2025	MeasuringWorth
REER	178 countries, monthly	~66	1960s-2026	Bruegel/Darvas
Commodities	~70 series (oil, metals, agriculture)	~66	1960-present	World Bank
Real wages	Labourers' real wage, ~40 countries	~180	1820-2000+	Clio Infra
Swedish FX	Exchange rates from 1534	492	1534-2026	Riksbank Hist.
Sovereign debt	Debt-to-GDP, 191 countries	215	1800-2015	IMF HPDD
Crisis indices	Banking/currency/debt crises	~216	1800-2016	Reinhart-Rogoff
FX regimes	De facto classifications	81	1940-2021	IRR
CFS FX/rates	Official + market exchange rates, interest rates	~450	~1500-1950	CFS

Source data schemas

`sources/memdb/`

memdb_spufford_medieval_exchange_rates.csv (13,197 rows) — Place, Date_start, Date_end, Type_of_Exchange, Currency_From, Amount_From, Currency_To, Amount_To, Notes, Source.

memdb_metz_currency_exchanges.csv (50,559 rows) — Place, Year, Coin_Ratio, Relationship, Value, Length_of_Series, Note.

`sources/clio_infra/`

All Clio Infra files share the same wide format: first column is year, remaining columns are country names. Values are yearly averages.

File	Unit	Period
`clio_infra_exchange_rates.csv`	Local currency per 1 USD	1500-2013
`clio_infra_exchange_rates_gbp.csv`	Local currency per 1 GBP	1500-2013
`clio_infra_inflation.csv`	Annual % change	1500-2010
`clio_infra_gold_standard.csv`	0/1 indicator	1800-2010
`clio_infra_bond_yield.csv`	Annual avg %	1727-2011
`clio_infra_govt_debt.csv`	% of GDP	1692-2010
`clio_infra_gdp_per_capita_compact.xlsx`	1990 Int'l GK dollars	1500-2016
`clio_infra_real_wages_compact.xlsx`	Subsistence ratios	1820-2000+

`sources/measuringworth/`

measuringworth_exchange_rates.csv — Wide format: year + 41 country columns. Values are foreign currency per 1 USD.

measuringworth_gold_prices.csv (769 rows) — year, British_price (GBP/oz, 1257-1945), london_price (GBP then USD), us_price (USD/oz, 1786-2025), newyork_price (USD/oz, 1791-2025), goldsilver_price (ratio, 1687-2025).

measuringworth_interest_rates.csv — UK+US short and long term rates, 1729-2025.

measuringworth_us_cpi.csv — US CPI index (avg 1982-84=100), 1774-2025.

measuringworth_dollar_pound.csv — USD per GBP, 1791-2025.

`sources/gold/`

gold_monthly_usd.csv — Date (YYYY-MM), Price (USD/troy oz). 1833-2025.

`sources/imf/`

imf_exchange_rates.csv (158K rows) — Date, Rate (LCU per USD), Currency, Frequency (M), Source, Country code, Country.

`sources/bis/`

BIS SDMX flat CSV format. xru/WS_XRU_csv_flat.csv.gz (1.5M rows): bilateral rates vs USD. eer/WS_EER_csv_flat.csv.gz (1.2M rows): nominal+real effective rates (2020=100).

`sources/fred/`

25 CSVs in daily/. Two columns: observation_date, rate. GBP/EUR/AUD/NZD quoted as USD-per-foreign; all others as foreign-per-USD. Missing values shown as ..

`sources/riksbank/`

riksbank_exchange_rates.csv (295K rows) — date, series_id (SEK[CURRENCY]PMI), value.

`sources/worldbank/`

worldbank_exchange_rates.csv — iso3, country, year, exchange_rate (LCU per USD). Via World Bank API.

`sources/worldbank_commodities/`

wb_commodity_prices_monthly.xlsx and wb_commodity_prices_annual.xlsx — ~70 commodities (crude oil Brent/WTI/Dubai, natural gas, coal, metals, agriculture) from the World Bank "Pink Sheet". 1960-present.

`sources/irr/`

Ilzetzki-Reinhart-Rogoff regime classifications. Rows are months (YYYY:MM), columns are countries. irr_regime_coarse.csv (1=peg to 6=dual market), irr_regime_fine.csv (1-15 scale), irr_anchor_master.csv (anchor currency), irr_unified_market_indicator.csv (0=unified, 1=parallel).

`sources/jst/`

jst_macrohistory.xlsx (2,719 rows, 59 cols) — Long format, one row per country-year. Key columns: year, country, iso, xrusd, cpi, stir, ltrate, peg, crisisJST. 18 countries, 1870-2017.

`sources/boe/`

boe_millennium.xlsx (26 MB, 90+ sheets) — UK data. Key sheets: A33 ($/£ from 1791), M14 (monthly bilateral rates 1963+), M15 (monthly $/£ 1791-2015), A31 (interest rates), D1 (daily Bank Rate).

`sources/gmd/`

gmd_exchange_rates.csv (57K rows) — ISO3, countryname, year, USDfx (LCU per USD), REER.

`sources/freegold/`

freegold_prices.csv — date, price. Gold prices 1258-2025 (GBP before 1791, USD after). freegold_silver_prices.csv — Silver prices 1688-2025. freegold_gold_silver_ratio.csv — Gold/silver ratio 1258-2025.

`sources/lbma/`

lbma_gold_daily.csv (14.5K rows) — date, gold_pm_usd, gold_pm_gbp, gold_pm_eur. Daily PM fix from 1968. lbma_silver_daily.csv (14.7K rows) — Same for silver.

`sources/schmelzing/`

schmelzing_real_interest_rates.xlsx (2.1 MB, 9 sheets) — Real interest rates 1311-2018 for Italy, UK, Netherlands, Germany, France, Spain, Japan, US. Key sheet: "II. Headline series" (719 rows x 35 cols).

`sources/maddison/`

maddison_gdp_per_capita.csv (21.6K rows) — entity_code, entity_name, year, gdp_per_capita. 178 countries, 1 CE-2022. Via OWID API (Maddison Project Database 2023).

`sources/allenunger/`

973 tab-delimited files, one per city-commodity pair (e.g. Amsterdam_Wheat.tab, London_Coal.tab). Columns: Commodity, Variety, Market, Original Measure, Standard Measure, Original Currency, Standard Currency, Year, Original Value, Standardized Value, Notes, Sources. Prices standardized to silver grams per litre. 1260-1914.

`sources/pwt/`

pwt.xlsx (6.3 MB) — Penn World Table 10.0/11.0. 185 countries, 1950-2023. Key variables: xr (exchange rate), pl_gdpo (price level of GDP), rgdpe (real GDP).

`sources/bruegel/`

REER_database_ver*.xls (9 MB) — Darvas/Bruegel real effective exchange rates. 178 countries, monthly, various start dates (many from 1960s). Nominal and CPI-based REER indices.

`sources/imf_hpdd/`

imf_hpdd_debt_gdp.csv (9.6K rows) — country, indicator, year, value. Gross government debt as % of GDP, 191 countries, 1800-2015. Via DBnomics mirror of IMF HPDD.

`sources/cfs/`

Center for Financial Stability Historical Financial Statistics. 8 files (~31 MB total). Key files: cfs_official_exchange_rates.xlsb (daily official FX rates), cfs_market_exchange_rates.xlsb (daily/monthly market FX), cfs_interest_rates.xlsb (interest rates), cfs_general_tables.xlsx (summary tables). Coverage: ~1500-1950. Note: .xlsb files require pyxlsb library to read.

`sources/riksbank_hist/`

Riksbank Historical Monetary Statistics, 13 Excel files across 3 volumes. Vol I (1277-2008): exchange rates, CPI, wages. Vol II (1620-2012): GDP, stocks/bonds, money supply, Riksbank balance sheet. Vol III (1420-2020): bonds. Key files: vol1_ch3_middle_ages.xls (earliest data, from 1277), vol1_ch4_exchange_rates_1534_1803.xls, vol1_ch8_cpi.xls.

`sources/reinhart_rogoff/`

12 Excel files from Reinhart & Rogoff. Key files: rr_regime_classification.xlsx (FX regime, annual), rr_anchor_currency_1946_2016.xlsx (anchor currency), rr_total_public_debt_gdp.xls (debt/GDP), rr_inflation_annual.xls, rr_bcdi_crisis_index.xls (banking/currency/debt/inflation crisis indicator), rr_gold_standard_dates.xlsx. Coverage: ~1800-2016, ~70 countries.

Derived data

Normalized (`data/derived/normalized/`)

File	Description
`yearly_unified_panel.csv`	243 countries, 1500-2025 (MW + CI + GMD with source priority tag)
`yearly_unified_wide.csv`	Same, year x country matrix
`fred_daily_normalized.csv`	23 FRED pairs, foreign-per-USD convention
`fred_daily_normalized_wide.csv`	Same, date x currency matrix

Analysis (`data/derived/analysis/`)

File	Description
`daily_log_returns.csv`	Daily log returns, 23 currencies, 271K obs
`yearly_log_returns.csv`	Annual log returns, 40 currencies (MeasuringWorth only)
`daily_volatility_stats.csv`	Vol, kurtosis, skew, 3-sigma tail counts per currency
`yearly_volatility_stats.csv`	Same at annual frequency for 40 countries
`daily_rolling_volatility.csv`	252-day rolling annualized vol, 231K obs
`daily_correlation_matrix.csv`	Pairwise Pearson correlations (23x23)
`yearly_correlation_matrix.csv`	Pairwise Pearson correlations (min 30 shared years)
`yearly_regime_classification.csv`	IRR regime per country-year (194 countries, 1940-2019)
`regime_conditional_stats.csv`	Volatility and kurtosis by regime type
`yearly_gold_inflation.csv`	Gold inflation, purchasing power, CPI gap (243 countries, 1257-2025)
`monthly_gold_inflation.csv`	Monthly gold inflation and debasement (174 currencies, 1940-2025)
`daily_momentum_signals.csv`	3/6/12-month trailing momentum for 23 currencies (Jegadeesh-Titman 1993)
`daily_momentum_reversals.csv`	Momentum reversal signals: 12m vs 1m divergence (crash precursor)
`sigma_event_frequency.csv`	2-5 sigma event counts vs Gaussian expected (Monday Morning Macro)
`jst_asset_returns.csv`	Real returns, vol, kurtosis for equities, housing, bonds, bills (Jordà et al. 2019)
`stock_bond_correlation.csv`	20-year rolling stock-bond correlation, 18 countries (Artemis 2020)

TODO

Manual data sources (not yet automatable)

Reinhart-Rogoff parallel/black-market exchange rates from carmenreinhart.com. Unique dataset — links broken from site migration, requires manual browser download.
Energy Institute Statistical Review from energyinst.org. Oil prices from 1861. Cloudflare-blocked — requires browser download.
MEMDB medieval exchange rates from memdb.libraries.rutgers.edu. Spufford (1106-1500) and Metz (1350-1800). No export API — requires web scraping.
Global Macro Database from globalmacrodata.com. 243 countries, USDfx + REER. Email-gated download.
NBER Macrohistory Database from nber.org. 2,510 series of pre-WWI/interwar data. Available via FRED individual series or NBER directory scrape.

Pipeline improvements

Integrate new sources into build.py (Schmelzing, Maddison, Allen-Unger, Bruegel REER, WB commodities, PWT, LBMA silver, interest rates, CPI, real wages)
Add silver purchasing power analysis (parallel to gold inflation)
Add real exchange rate computation using CPI and Bruegel REER data
Cross-validate overlapping series (e.g. MeasuringWorth gold vs FreeGoldAPI vs LBMA)

References

Allen, R.C. & Unger, R.W. "Global Commodity Prices Database." International Institute of Social History.
Bolt, J. & van Zanden, J.L. (2024). "Maddison style estimates of the evolution of the world economy. A new 2023 update." Journal of Economic Surveys, 38(4), 1507-1545.
Clark, G. "What Were the UK Earnings and Prices Then?" MeasuringWorth.
Darvas, Z. (2012). "Real effective exchange rates for 178 countries: A new database." Bruegel Working Paper 2012/06.
Denzel, M.A. (2010). Handbook of World Exchange Rates, 1590-1914. Ashgate/Routledge.
Feenstra, R.C., Inklaar, R. & Timmer, M.P. (2015). "The Next Generation of the Penn World Table." AER, 105(10), 3150-3182.
Ilzetzki, E., Reinhart, C.M. & Rogoff, K.S. (2019). "Exchange Arrangements Entering the 21st Century." QJE, 134(2), 599-646.
Jegadeesh, N. & Titman, S. (1993). "Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency." Journal of Finance, 48(1), 65-91.
Jorda, O., Knoll, K., Kuvshinov, D., Schularick, M. & Taylor, A.M. (2019). "The Rate of Return on Everything, 1870-2015." QJE, 134(3), 1225-1298.
Jorda, O., Schularick, M. & Taylor, A.M. (2017). "Macrofinancial History and the New Business Cycle Facts." NBER Macroeconomics Annual, 31(1), 213-263.
Metz, R. (1990). Geld, Wahrung und Preisentwicklung: der Niederrheinraum im europaischen Vergleich, 1350-1800. Frankfurt.
Officer, L.H. "What Was the Interest Rate Then?" MeasuringWorth.
Officer, L.H. "Dollar-Pound Exchange Rate From 1791." MeasuringWorth.
Officer, L.H. & Williamson, S.H. "The Price of Gold, 1257-Present." MeasuringWorth.
Officer, L.H. & Williamson, S.H. "The Annual Consumer Price Index for the United States, 1774-Present." MeasuringWorth.
Reinhart, C.M. & Rogoff, K.S. (2009). This Time Is Different: Eight Centuries of Financial Folly. Princeton University Press.
Schmelzing, P. (2020). "Eight centuries of global real interest rates, R-G, and the 'suprasecular' decline, 1311-2018." Bank of England Staff Working Paper No. 845.
Spufford, P. (1986). Handbook of Medieval Exchange. Royal Historical Society.
Rattray, S., Harvey, C.R. & Van Hemert, O. (2018). "Strategic Rebalancing." Journal of Portfolio Management, 44(4), 18-31.
Scowcroft, A. & Sefton, J. (2005). "Understanding Momentum." Financial Analysts Journal, 61(2), 64-82.
Thomas, R. & Dimsdale, N. (2017). "A Millennium of UK Data." Bank of England OBRA dataset.

Name		Name	Last commit message	Last commit date
.github/workflows		.github/workflows
charts		charts
data		data
notebooks		notebooks
scripts		scripts
tests		tests
.gitignore		.gitignore
CITATION.cff		CITATION.cff
LICENSE		LICENSE
Makefile		Makefile
README.md		README.md
build.py		build.py
flake.lock		flake.lock
flake.nix		flake.nix
quickstart.py		quickstart.py
quickstart_pandas.py		quickstart_pandas.py
requirements.txt		requirements.txt
validate.py		validate.py
visualize.py		visualize.py

Latest commit

History

Folders and files

Repository files navigation

forex-centuries

Data overview

Citation

Coverage

What the data shows

Fat tails are universal and persistent

The peg paradox

Volatility clusters by regime

Cross-currency correlations

Implications

Quickstart

Build pipeline

Testing

Interactive notebook

Updating source data

Charts

Project structure

Sources

Automated (update_sources.py --all)

Static (committed to repo, updated manually)

Longest series by asset type

Source data schemas

sources/memdb/

sources/clio_infra/

sources/measuringworth/

sources/gold/

sources/imf/

sources/bis/

sources/fred/

sources/riksbank/

sources/worldbank/

sources/worldbank_commodities/

sources/irr/

sources/jst/

sources/boe/

sources/gmd/

sources/freegold/

sources/lbma/

sources/schmelzing/

sources/maddison/

sources/allenunger/

sources/pwt/

sources/bruegel/

sources/imf_hpdd/

sources/cfs/

sources/riksbank_hist/

sources/reinhart_rogoff/