Sources & method

How it works

This is a static site. A script downloads every series, calculates what needs calculating and writes the pages; GitHub rebuilds and publishes it automatically on Monday to Saturday at 23:30 UTC, after the US futures markets and the Asian exchanges have closed. A single series that fails never blocks the site: it becomes "unavailable" and a warning appears at the top of the pages that use it. Prices are closing prices without dividends; rates are in % and changes in rates are shown in basis points (bp, 0.01%), changes in year-on-year inflation in percentage points (pp). This dashboard is for study purposes and is not financial advice.

Sources

The number of series is counted by main source. Every chart names its sources under the chart, with a link to the series on the source's website; the full list is in Series status.

Recession dates: NBER (series USREC on FRED) for the US; the CEPR for the euro area, copied by hand into the configuration because it has no API.

Freshness check

Every series is checked after each download. If its latest data point is older than the threshold for its frequency, a warning appears at the top of the pages that use it. It catches "frozen" series: the source answers without any error but has stopped adding data (for example when a dataset is retired and replaced by another). Days are counted from the end of the period of the latest data point, so the August value of a monthly series counts as fresh until 31 August.

A few series have their own threshold, always for a documented reason: BoK base rate (45 days) — the BIS publishes South Korea with about a month of delay (see Korea policy rate delay).

Right now 2 series are stale: Korea CPI inflation (latest data 1 Jul 2026), Japan CPI inflation (latest data 1 Jul 2026).

Annual data by country

Country data that changes only once or twice a year (GDP, inflation, government balance and debt and the IMF projections from the World Economic Outlook, WEO; unemployment, population and governance indicators from the World Bank) is not downloaded at every build. It sits in files in the repository (a "snapshot"): the WEO is refreshed by hand after each release, in April and October (the October 2026 edition is due in mid-October), and the World Bank files after its releases (July for the development indicators, September for the governance indicators). The daily build only reads the files. In the IMF snapshot every figure after the country's latest actual year is an estimate or a projection, and the site will draw it differently. The freshness check counts from the publication date: a warning appears if the IMF snapshot is more than 210 days old (six months plus about four weeks, so a new edition does not trigger a false alarm) or a World Bank snapshot more than 456 days (fifteen months). Details of each indicator are in Series status.

Fallback sources

Some series have a second source, used only if the main one fails. When it happens the chart footer says "(fallback)" and a notice appears on the page. A fallback can be a slightly different quantity (for example a spot price instead of a future, or a monthly average instead of a daily value), a different unit, or a series with a shorter history: the difference is written in the notice and in the table below. Right now every series uses its main source.

Calculated series

A calculated series (a difference or a ratio of two others) is computed only if all its inputs come from the same main source. If even one input has switched to its fallback, the calculated series becomes "unavailable" with a warning that explains why: spot and futures prices, different sources or different units are never mixed (for example Brent spot from FRED minus WTI futures from Yahoo).

What changed this week

The Overview lists the biggest weekly moves compared with each series' own normal volatility, so that a 20 bp move in a Treasury yield and a 5% move in oil can be compared. For each daily or weekly series flagged for this purpose, the site takes the change over the last 7 days and divides it by the typical size of 7-day changes in the 3 years before that week (the standard deviation, from at least 100 observations). The result is shown as "×normal": 3.0× means three times a normal week. Rates and spreads are measured in basis points, prices and indices in %.

Only moves of at least 2× are listed, at most one series per group (all US Treasury yields count as one, for example, because they usually move together) and at most 5 rows. If nothing reaches the bar, the box says "A quiet week".

Why 2× and not a lower bar? If weekly moves were normally distributed, about 13% of series would exceed 1.5× in an ordinary week purely by chance, so the list would almost always be full and would lose its meaning. At 2× that share falls to about 5%, so the list fills up mainly when something unusual really happens, and "A quiet week" becomes useful information. Real markets have fatter tails and many series move together, so in a turbulent week the share can be much higher: on the latest build, 0 of 54 series exceeded 2×.

A score is not a probability. It says how many times larger than a typical week the move was, not how rare it is. The VIX, for example, moves a lot even in normal weeks, so it rarely reaches the bar. Policy rates are left out because they change in steps, monthly series because they have no weekly move, and any series with fewer than 100 past observations or an out-of-date last value.

Known limits

The technical notes that used to sit under the charts. Under each chart you will find links to the notes that apply to it.

Chart conventions

US recession bands (NBER)

Grey bands: US recessions, dated by the NBER (series USREC on FRED). They are drawn only on charts of a single economy, where they mean something; charts that compare several countries have no bands. Used on: 10Y-2Y and 10Y-3M spreads, 10Y real yield and 10Y breakeven (US), US yields by maturity over time, Headline CPI, core CPI and core PCE (% y/y), Investment Grade and High Yield OAS (ICE BofA), Baa (Moody's) spread vs 10Y Treasury, Concentration: S&P 500 Equal Weight / S&P 500, VIX (expected volatility of the S&P 500), Broad dollar index (Jan 2006 = 100), WTI and Brent crude oil, Henry Hub natural gas (US), TTF natural gas (Europe), Gold, Silver, Copper, Aluminium (COMEX), Wheat, Corn, Gold vs 10-year US real yield (TIPS), Copper/gold ratio vs 10-year Treasury, WTI crude oil vs 10-year expected inflation (breakeven), Brent − WTI spread, Unemployment rate

Euro-area recession bands (CEPR)

Grey bands: euro-area recessions dated by the CEPR. There is no API for them: the dates are written by hand in config.yaml and must be checked manually when the CEPR announces a new peak or trough. The OECD indicator on FRED stopped in August 2022 and the CEPR website blocks automatic requests. Used on: Euro-area AAA curve (Bund proxy): 2 and 10 years, HICP headline and core (% y/y), BTP-Bund and OAT-Bund 10-year spreads (MONTHLY data), All euro-area government bonds minus AAA, 10 years (daily)

Curve inversion (red area)

Red area: inverted curve, that is the spread between a long and a short maturity is below zero. An inverted curve has often preceded recessions. Used on: Curve slope 10Y-2Y: US vs euro area, 10Y-2Y and 10Y-3M spreads

Fed 2% target

Dashed line: the Fed's 2% inflation target (measured on the PCE price index). Used on: Headline CPI, core CPI and core PCE (% y/y)

ECB 2% target

Dashed line: the ECB's 2% inflation target. Used on: HICP headline and core (% y/y)

BoJ 2% target

Dashed line: the BoJ's 2% inflation target. Used on: Headline, core and core-core CPI (% y/y)

BoK 2% target

Dashed line: the BoK's 2% inflation target.

2% reference line

Dashed line: 2%, the inflation target of many central banks. Monthly data. Used on: CPI inflation compared (% y/y)

Zero line (deflation)

Dashed line: zero. Below it, consumer prices are falling (deflation).

Base 100 charts

In "base 100" charts every line is rebased to 100 at the start of the period chosen with the 1Y / 5Y / 10Y / Max buttons, but never earlier than the first date on which all the visible lines exist. The rebasing is done by the page, so changing the period changes every value. Used on: Equity indices compared (base 100), US equity indices (base 100), European equity indices and banks (base 100), Currencies against the dollar (base 100)

Rates, curves and inflation

Fed and ECB rates

The effective Fed funds rate (series DFF on FRED, daily) is the average rate banks actually pay each other overnight: it sits inside the Fed's target range. The ECB deposit facility rate (ECB Data Portal, with a FRED fallback) is the rate banks earn on money deposited at the ECB, today the ECB's main tool. The BIS series in the chart above report the midpoint of the Fed's target range instead. Used on: Fed funds effective rate and ECB deposit rate

How the curve snapshot is dated

"Today" is the latest day on which every maturity has a value. "1 month ago" and "1 year ago" use the last day on or before that date on which every maturity has a value. Used on: Treasury curve: today, 1 month ago, 1 year ago, Euro-area AAA curve (Bund proxy): today, 1 month ago, 1 year ago

AAA curve as a Bund proxy

The ECB estimates a yield curve for euro-area government bonds rated AAA (Germany, the Netherlands and a few others), daily from September 2004. It is used as an approximation of German Bund yields, because daily yields on German Bunds alone are not freely available. The curve of all euro-area government bonds ("G_N_C") has the same history. Used on: Euro-area AAA curve (Bund proxy): today, 1 month ago, 1 year ago, Curve slope 10Y-2Y: US vs euro area, Euro-area AAA curve (Bund proxy): 2 and 10 years

TIPS and breakeven history

Real yields on TIPS and the breakeven inflation rate (nominal minus real yield) are available from 2003 only. Used on: 10Y real yield and 10Y breakeven (US)

HICP core definition

Core = excluding energy, food, alcohol and tobacco. The headline and core series come from the new ECB dataset "HICP" (code 4D0): the old dataset "ICP" stopped in December 2025 and still answers without errors, which is why the freshness check exists. Used on: HICP headline and core (% y/y)

Policy rates compared

All series come from the BIS. Fed: midpoint of the target range; ECB: deposit facility rate; BoJ: target overnight rate; BoK: base rate. WARNING: for China it is the 1-year Loan Prime Rate (LPR), a bank lending rate, not an overnight rate like the others: it tends to sit higher and is not directly comparable. Korea is about a month behind. Used on: Policy rates compared

China Loan Prime Rate

Source: BIS, daily data. From 20 Aug 2019 it is the 1-year LPR; before that the BIS uses the official 1-year lending rate. Used on: Policy rates compared

Korea policy rate delay

Source: BIS, daily data. The BIS publishes Korea with about a month of delay: for this series the stale-data warning triggers after 45 days instead of 10. Used on: Policy rates compared

10-year yields compared

US: Treasuries (FRED, daily); euro area: the ECB AAA curve as a Bund proxy (daily); Japan: JGBs from the Ministry of Finance (daily); South Korea: OECD MONTHLY average. China is not included: there is no free, up-to-date source. Daily yields on German Bunds alone are not freely available either. Used on: 10-year government bond yields

Korea 10-year bond is monthly

Monthly data (monthly average, source OECD via FRED), with about a month of delay: the daily data is not freely available. Used on: 10-year government bond yields

JGB history and fallback

The 30-year starts in 1999, the 10-year in 1986. If the Ministry of Finance does not respond the 10-year alone falls back on the OECD monthly average (FRED). Used on: JGB yields at 2, 10 and 30 years

Japan CPI sources

All three series come from the Statistics Bureau of Japan (via DBnomics): the annual change is calculated from the index. In the Global comparison Japan uses BIS data instead, so that every country has the same source. "Core" excludes fresh food; "core-core" also excludes energy. Used on: Headline, core and core-core CPI (% y/y)

Credit and sovereign spreads

Baa spread has a long history

A long-history series (from 1986): useful to compare past crises. Used on: Baa (Moody's) spread vs 10Y Treasury

ICE BofA OAS start in 2023

ICE BofA OAS series on FRED only start in October 2023: for licensing reasons FRED publishes only the last few years. For the long history of US credit see the Baa vs 10Y Treasury spread. Used on: High-yield spread: US vs euro area, Investment Grade and High Yield OAS (ICE BofA)

No euro Investment Grade spread

The euro Investment Grade spread is not included: there is no free source. For euro High Yield, FRED publishes the ICE BofA series only from October 2023. Used on: High-yield spread: US vs euro area

BTP-Bund and OAT-Bund are monthly

Monthly data: the monthly average of the 10-year yields the ECB uses for its convergence criteria (Germany from 1990, Italy from 1991, France from 1986), published with about a month's delay. Daily yields of individual countries are not freely available. Used on: BTP-Bund and OAT-Bund 10-year spreads (MONTHLY data)

All euro-area bonds minus AAA

10-year yield of the ECB curve for all euro-area government bonds minus that of the AAA curve: a daily indicator of the average risk premium of non-AAA countries (Italy, France, Spain...). Data from 2004. Used on: All euro-area government bonds minus AAA, 10 years (daily)

Equities

Prices without dividends

All equity prices are closing prices without dividends (not "adjusted"), for consistency between indices and ETFs. ETFs that pay dividends (ACWI, the European banks ETF) show them in the price as small drops. Used on: Equity indices compared (base 100), US equity indices (base 100), European equity indices and banks (base 100)

Index levels are daily closes

Index levels (in points, or the price in local currency for ETFs) are daily closing prices from Yahoo Finance, an unofficial source. The last candle is dropped if the trading session has not finished yet. Used on: Equity indices compared (base 100)

What the VIX is

The VIX is the CBOE Volatility Index: the volatility of the S&P 500 that investors expect over the next 30 days, calculated from option prices (series VIXCLS on FRED, daily close). It is quoted in points, not in currency: a value of 20 corresponds to an expected move of roughly 20% over a year, or about 6% over a month. Used on: VIX (expected volatility of the S&P 500)

US index definitions

S&P 500: 500 large companies; Nasdaq 100: the 100 largest non-financial companies on the Nasdaq (a lot of technology); Russell 2000: small companies. Used on: US equity indices (base 100)

S&P 500 concentration ratio

Ratio (x100) between the S&P 500 with every stock at the same weight and the S&P 500 weighted by market capitalisation. When it falls, the market is concentrating on a few giants: the "average" stock does worse than the index. When it rises, the rally broadens. What matters is the direction, not the level (it depends on the indices' start date, Dec 2006). Price indices, without dividends. The ETFs RSP/SPY (from 2003) are not needed: the ratio of the indices is almost identical. Used on: Concentration: S&P 500 Equal Weight / S&P 500

DAX is the price index

The DAX K (Kursindex, a price index) is used. The "official" DAX index (^GDAXI) reinvests dividends and would not be comparable with the other indices. Yahoo's price index starts in March 2013, so with "Max" the euro-area equity chart starts in 2013, when every line exists. Used on: European equity indices and banks (base 100)

European banks are an ETF

"European banks" is the iShares STOXX Europe 600 Banks ETF (Xetra, ticker EXV1), in euro: it also includes banks outside the euro area (e.g. the UK, Switzerland) and is NOT the Euro Stoxx Banks, which is not available on Yahoo. Xetra's latest price is often from the previous day. Used on: European equity indices and banks (base 100)

Local currency vs USD

With "Local currency" performance ignores exchange rates (the S&P 500 and MSCI ACWI are already in dollars); with "In USD" each index is divided by the Yahoo Finance exchange rate of the same day, so it includes the currency effect. If an exchange rate is unavailable the USD line becomes "unavailable" instead of mixing sources. Markets and exchange rates do not close at the same time: small one-day differences. Used on: Equity indices compared (base 100)

MSCI ACWI is an ETF

Dashed line: MSCI ACWI, the world benchmark index, represented by the iShares ACWI ETF (in dollars, from March 2008) because the index itself is not on Yahoo. The ETF pays out dividends, which show up in the price as small drops. Used on: Equity indices compared (base 100)

CSI 300 is an ETF

The CSI 300 is represented by an ETF (ticker 510300 on the Shanghai exchange, price in yuan), not by the index, because Yahoo does not provide the history of the index (it returns a single candle). The price follows the index but is not equal to its level; data start in 2012, so with "Max" every comparison chart starts on the same date. Used on: Equity indices compared (base 100)

Hang Seng and the HK dollar

Hang Seng: the Hong Kong dollar is pegged to the US dollar, so the conversion to USD matters little. Used on: Equity indices compared (base 100)

Currencies

EUR/USD ECB reference rate

ECB reference rate, taken every working day at 14:15 (Frankfurt time). Up = the euro strengthens. There is no FRED fallback with the same quotation direction for the euro. Used on: EUR/USD (dollars per 1 euro)

USD/JPY quotation

Up = the yen weakens. If Yahoo does not respond FRED is used (the Fed's daily rate, with a delay of a few days). Used on: USD/JPY (yen per 1 dollar)

USD/CNY quotation

Onshore rate. Up = the yuan weakens. If Yahoo does not respond FRED is used. Used on: USD/CNY (yuan per 1 dollar)

USD/KRW quotation

Up = the won weakens. If Yahoo does not respond FRED is used. Used on: USD/KRW (won per 1 dollar)

Broad dollar index

The Federal Reserve's nominal broad dollar index (series DTWEXBGS on FRED), January 2006 = 100: the dollar against the currencies of the United States' main trading partners, weighted by trade. It is published weekly with a few days of delay and starts in 2006. Used on: Broad dollar index (Jan 2006 = 100)

Currencies base 100

Above 100 = the currency has strengthened against the dollar since the start of the period; below 100 = it has weakened. The yen, yuan and won are quoted as "currency per dollar" and are flipped here, so every line reads the same way. Euro: ECB rate at 14:15; the other currencies: Yahoo Finance close. With "Max" the chart starts on the first date when all the currencies exist. Used on: Currencies against the dollar (base 100)

Commodities

Continuous futures

Yahoo Finance continuous futures: the contract with the nearest expiry is always followed and, when it expires, the next one takes over. On those days small price jumps can appear that are not real market moves (for example the November Brent contract expires on the last working day of September). Shown only when the data really comes from Yahoo, not from the FRED fallback. Used on: WTI and Brent crude oil, Henry Hub natural gas (US), TTF natural gas (Europe), Gold, Silver, Copper, Aluminium (COMEX), Wheat, Corn, Gold vs 10-year US real yield (TIPS), Copper/gold ratio vs 10-year Treasury, WTI crude oil vs 10-year expected inflation (breakeven), Brent − WTI spread

WTI below zero in 2020

On 20 Apr 2020 the WTI future closed below zero (-$37): at expiry nobody had room left to store the oil. Negative values are not discarded. Used on: WTI and Brent crude oil

TTF history

Data on Yahoo from 2017. Used on: TTF natural gas (Europe)

Aluminium is a thin future

A thinly traded COMEX future (the world benchmark is the London Metal Exchange, which is not free): the price is consistent with the LME one but can move in jumps. Data on Yahoo from 2014. Used on: Aluminium (COMEX)

Gold vs real yield

Gold pays no interest: when the real yield rises, holding it "costs" more and the price tends to fall. That is why the real-yield axis is INVERTED (high values at the bottom): if the link holds, the two lines rise and fall together. Since 2022 the link has weakened because of gold purchases by central banks. Gold and silver have no free fallback source: if Yahoo does not respond the charts stay empty. Used on: Gold vs 10-year US real yield (TIPS)

Copper/gold ratio

Copper (a cyclical metal) divided by gold (a safe haven), multiplied by 1000: it rises when the market expects more growth and tends to move together with 10-year yields. Calculated only if both prices come from Yahoo. Used on: Copper/gold ratio vs 10-year Treasury

Oil vs breakeven inflation

The price of energy weighs heavily on the inflation the market expects (breakeven = nominal yield minus real yield): the two series tend to move together. Used on: WTI crude oil vs 10-year expected inflation (breakeven)

Brent-WTI spread

Difference between the two oil futures. Calculated only if both prices come from Yahoo: the FRED fallbacks are spot prices and are never mixed with futures (at the end of September 2026 Brent spot was about $11 above the future). Brent is an ICE contract that expires before WTI: on expiry days the spread can jump. Used on: Brent − WTI spread

Country coverage and data not included

China data not included

Not included for lack of free, up-to-date sources: 10-year yield, 5-year LPR, PPI (the National Bureau of Statistics series on DBnomics stopped in Feb 2026; FRED/OECD stop in 2022-2023). FRED/OECD CPI for Japan and China are frozen (2021 and 2025): never use them.

Korea data not included

Not included: 3-year yield (the daily data is only available through the Bank of Korea ECOS API, which requires a registration with a Korean phone number). The 10-year yield is monthly only.

Country pages (annual data by country)

IMF estimates and projections

IMF data comes from the World Economic Outlook (WEO), published in April and October and revised every time. For each country and indicator the IMF flags the "latest actual year", the last year with a figure reported by national statistics; later years are IMF estimates (years already past but not yet confirmed) and projections, up to 2031. They are drawn dashed (lines) or lighter (bars) and the tooltip says so. The latest actual year differs between countries and indicators, so the year is always written next to the value. A few countries have no projections at all (the IMF data stop at the last actual year), and for a few the data stop long before: the page says so under the chart. The projections of the years further ahead are less certain than the nearer ones. Used on: Real GDP growth, Nominal GDP, GDP per head, PPP, Inflation, Government balance, Government debt, Current account balance, IMF outlook

Fiscal years in the IMF data

For some countries the IMF reports fiscal years instead of calendar years, for some or all indicators (public finances for about thirty countries, national accounts and prices for fewer). A fiscal year is shown under the year in which it starts: for India, 2024 means the fiscal year 2024/25 (April 2024 to March 2025). The note appears under the charts concerned. Fiscal years can differ in length from country to country, so comparisons between countries are approximate. Used on: Real GDP growth, Nominal GDP, GDP per head, PPP, Inflation, Government balance, Government debt, Current account balance, IMF outlook

ILO modelled estimates (unemployment and employment)

The unemployment rate and the employment ratio on the country pages are modelled estimates of the International Labour Organization (ILO), taken from the World Bank. They are built with a statistical model that uses labour-force surveys where they exist and fills the gaps, with the same definition for every country. They are therefore comparable across countries but they are not the figures published by each national statistical office, which can differ slightly. The employment ratio is the share of the population aged 15 or over that has a job, so it also depends on retirement ages and on the participation of women and young people. Used on: Unemployment rate (ILO modelled estimate), Employment ratio, 15+ (ILO modelled estimate)

Control of corruption score (WGI)

The World Bank's Worldwide Governance Indicators combine many surveys and expert assessments of how well public power is kept from private gain. Since the 2025 update (WGI 2.0) the result is a score on an absolute 0-100 scale anchored to fixed reference points (it replaces the old percentile rank, which is no longer published), recalculated back to 1996. It measures perceptions, not corruption itself. The band is the 90% confidence interval: it is wide for countries covered by few sources, and differences smaller than the band are not meaningful. Data exist for 1996, 1998 and 2000 and every year from 2002: the earlier years are shown as separate points. Used on: Control of corruption, score (0-100)

Countries on the country pages

The country pages cover every country and territory in the IMF World Economic Outlook or in the World Bank data (aggregates such as "World" or "Euro area" are not included). Some small countries and territories are only in the World Bank data: for them GDP, growth and GDP per head come from the World Bank, with no projections, and the pages have no inflation, public finances or current account. Taiwan is only in the IMF data (the name is the IMF's) and has no World Bank indicators. The IMF codes for Kosovo and for the West Bank and Gaza are replaced by the ISO codes used by the World Bank. The two sources are never mixed in the same chart.