About the evidence

Sources & Methods

Where the observations come from, what Pulse changes, and where the evidence stops.

Source register reviewed 18 Sept 2026

Two evidence bases, kept separate

Market briefings use externally published economic and reference-rate observations. PineConnector’s monthly research describes aggregated historical instructions received by its bridge. The two datasets are not combined to estimate market participation, trading outcomes or the size of PineConnector’s business.

Read the methodology for the existing network research. The source entries below describe the external data used in the current market briefing.

European Central Bank

Euro foreign exchange reference rates

Used for EUR, JPY. Included in the latest validated snapshot.

Indicators
  • EUR/USD reference rate · USD per EUR · no seasonal adjustment
  • EUR/JPY reference rate · JPY per EUR · no seasonal adjustment
  • USD/JPY derived reference rate · JPY per USD · no seasonal adjustment
Retrieved by Pulse
Unknown
Publication cadence
Daily on ECB publication days. No weekend or closure-day observation is invented.
Pulse calculations
EUR/USD uses USD per EUR directly. USD/JPY divides JPY per EUR by USD per EUR on the same date. Changes compare observed reference rates.
Important limits
Informational reference rates, not live quotes, executable prices, exchange closes or intraday candles.
Revisions
A changed historical observation creates a new snapshot; earlier briefings retain their original values.
Reuse & attribution
Reuse with accurate attribution and disclosure of modifications. Pulse calculates cross-rates and percentage changes; these are not ECB calculations.

Board of Governors of the Federal Reserve System

Federal funds target range

Used for USD. Included in the latest validated snapshot.

Indicators
  • Federal funds target lower bound · percent · no seasonal adjustment
  • Federal funds target upper bound · percent · no seasonal adjustment
Retrieved by Pulse
Unknown
Publication cadence
Updates when the target range changes; meeting dates come from the official FOMC calendar.
Pulse calculations
Preserve the lower and upper endpoints separately. Do not substitute a midpoint for the official range.
Important limits
The target range is not the effective federal funds rate or a market-implied forecast. Table dates are effective dates.
Revisions
Corrections in the official table are detected as revisions.
Reuse & attribution
Board material may be copied and distributed unless otherwise identified. Pulse cites the Board; no third-party artwork is reused.
Release calendars
Official schedule ↗

European Central Bank

ECB deposit facility rate

Used for EUR. Included in the latest validated snapshot.

Indicators
  • Deposit facility rate · percent · no seasonal adjustment
Retrieved by Pulse
Unknown
Publication cadence
Policy decisions and their stated effective dates.
Pulse calculations
Read the deposit facility column and its effective date. Future-effective changes are shown separately.
Important limits
The deposit facility rate is used for this comparison; it is distinct from the main refinancing and marginal lending rates.
Revisions
Historical corrections create new observations without changing earlier publications.
Reuse & attribution
Reuse with attribution; Pulse's comparisons and presentation are identified as modifications.
Release calendars
Official schedule ↗

Bank for International Settlements

Japan policy-rate history

Used for JPY. Included in the latest validated snapshot.

Indicators
  • Policy rate, BIS observation · percent · no seasonal adjustment · Source identifier: WS_CBPOL:D.JP
Retrieved by Pulse
Unknown
Publication cadence
Daily observations are normally released around mid-week. The last observed date is shown; this is not a real-time central-bank announcement service.
Pulse calculations
Select Japan daily policy-rate observations, preserving reported dates and units. No implied future policy path is computed.
Important limits
BIS compiles the series from central-bank reporting. It can lag a new announcement. Historical instrument changes follow the BIS compilation notes.
Revisions
BIS corrections create a new snapshot. Earlier briefing values stay unchanged.
Reuse & attribution
BIS statistics may be included in commercial publications with attribution and no additional charge for the statistics. Pulse provides this section freely and does not imply BIS endorsement.

US Bureau of Labor Statistics

US inflation and employment

Used for USD. Included in the latest validated snapshot.

Indicators
  • All-items CPI, annual change · percent yoy · not seasonally adjusted · Source identifier: CUUR0000SA0
  • Unemployment rate · percent labour force · seasonally adjusted · Source identifier: LNS14000000
  • Nonfarm payroll monthly change · thousand jobs change · seasonally adjusted · Source identifier: CES0000000001
Retrieved by Pulse
Unknown
Publication cadence
Monthly CPI and Employment Situation releases; calendars are checked separately.
Pulse calculations
Headline CPI annual change is calculated from the same index series twelve months apart. Payroll change is the difference between adjacent monthly levels, in thousands.
Important limits
CPI is not seasonally adjusted for the annual comparison. Payroll and unemployment series are seasonally adjusted; they come from different surveys.
Revisions
Latest API vintages may revise prior values. Preserve as-retrieved snapshots and identify revised periods.
Reuse & attribution
BLS statistical material is public domain. Source attribution is provided; copyrighted images are not used.
Release calendars
Official schedule ↗

US Bureau of Economic Analysis

US real gross domestic product

Used for USD. Included in the latest validated snapshot.

Indicators
  • Real GDP growth, annualised quarterly · percent qoq annualised · seasonally adjusted
Retrieved by Pulse
Unknown
Publication cadence
Quarterly estimates, subsequent estimates and annual revisions; next release date is read from BEA.
Pulse calculations
Retain the official annualised rate. A separately labelled non-annualised comparison may use ((1 + rate/100)^(1/4) - 1) × 100.
Important limits
The headline US quarterly growth rate is seasonally adjusted and annualised. It is not directly comparable with an ordinary quarter-on-quarter rate.
Revisions
Advance, second and third estimates and later revisions remain distinct snapshot vintages.
Reuse & attribution
BEA information is public domain unless marked otherwise. Source links and estimate stage are provided.
Release calendars
Official schedule ↗

Eurostat

Euro-area inflation, labour and growth

Used for EUR. Included in the latest validated snapshot.

Indicators
  • All-items HICP, annual change · percent yoy · not seasonally adjusted · Source identifier: prc_hicp_minr:M.RCH_A.TOTAL.EA21
  • Unemployment rate · percent labour force · seasonally adjusted · Source identifier: une_rt_m:M.SA.TOTAL.PC_ACT.T.EA21
  • Real GDP growth, quarterly · percent qoq · seasonally and calendar adjusted · Source identifier: namq_10_gdp:Q.CLV_PCH_PRE.SCA.B1GQ.EA21
Retrieved by Pulse
Unknown
Publication cadence
Monthly HICP and unemployment; quarterly GDP with subsequent estimates.
Pulse calculations
Read explicitly filtered series: all-items annual HICP, total seasonally adjusted unemployment, and seasonally/calendar-adjusted real quarterly GDP growth.
Important limits
Use the EA21 geography consistently. HICP follows ECOICOP version 2; the legacy classification stopped updating after 2025. A dataset-update timestamp is not every observation's release time.
Revisions
Retain Eurostat status flags and snapshot vintages; revised historical values never change an earlier briefing.
Reuse & attribution
Statistical reuse is permitted with acknowledgement, subject to source-specific exceptions. Pulse selects and presents the data; Eurostat is not responsible for these modifications.
Release calendars
Official schedule ↗

Statistics Bureau of Japan

Japan consumer price index

Used for JPY. Included in the latest validated snapshot.

Indicators
  • Nationwide all-items CPI, annual change · percent yoy · not seasonally adjusted
Retrieved by Pulse
Unknown
Publication cadence
Monthly nationwide CPI. Tokyo preliminary CPI is a separate population and is not substituted.
Pulse calculations
Discover the current nationwide monthly CSV, select all items and calculate twelve-month change only within one consistent index base. Reconcile the latest rounded result with the official summary.
Important limits
Nationwide all-items CPI uses the current published base. A base revision must not splice incompatible index levels. Japanese-language source labels are translated for display.
Revisions
Base revisions and corrected historical indices create new snapshots and explicit definition changes.
Reuse & attribution
Numerical statistical data may be used freely. Attribute the Statistics Bureau and identify Pulse calculations and translation; no agency endorsement is implied.
Release calendars
Official schedule ↗

Statistics Bureau of Japan

Japan Labour Force Survey

Used for JPY. Included in the latest validated snapshot.

Indicators
  • Unemployment rate · percent labour force · seasonally adjusted
Retrieved by Pulse
Unknown
Publication cadence
Monthly basic tabulation; use its official release calendar.
Pulse calculations
Read labelled monthly adjusted rates and their reference months from the official summary table.
Important limits
Use the seasonally adjusted unemployment rate. Annual averages and unadjusted rates are not interchangeable with monthly adjusted observations.
Revisions
Seasonal factors and population benchmarks may revise past rates. Preserve each publication's vintage.
Reuse & attribution
Numerical statistical data may be used freely with the Labour Force Survey source identified. English presentation is by Pulse.
Release calendars
Official schedule ↗

Cabinet Office of Japan, Economic and Social Research Institute

Japan quarterly real GDP

Used for JPY. Included in the latest validated snapshot.

Indicators
  • Real GDP growth, quarterly · percent qoq · seasonally adjusted
Retrieved by Pulse
Unknown
Publication cadence
First and second quarterly estimates and subsequent revisions.
Pulse calculations
Discover the current estimate's official time-series menu and select the real seasonally adjusted quarterly growth CSV by its documented table label.
Important limits
Real, seasonally adjusted quarter-on-quarter growth is selected. Nominal, annualised, calendar-year and fiscal-year measures remain separate.
Revisions
Each newly retrieved estimate is preserved; prior briefing data is immutable.
Reuse & attribution
Numerical data may be used freely. Attribute the Cabinet Office and identify edits or transformations without implying government authorship of Pulse analysis.
Release calendars
Official schedule ↗

External event schedule

Forex Factory economic calendar

The economic calendar uses Forex Factory’s official weekly export, separately from the macro briefing and PineConnector instruction research.

Coverage
Only the available weekly schedule. No historical API or actual release results are supplied. Forecasts and previous values retain the source’s strings and units.
Times & impact
Offset-bearing scheduled times are converted to the selected time zone. The export does not reliably mark all-day or tentative events. Holidays and unconfirmed dates remain separate from the daily agenda. Impact labels come from Forex Factory.
Updates
Source checks are scheduled every two hours. Checks and publication can be delayed. Successful checks and source modification dates are shown separately. A failed check retains the last valid, dated schedule; an ended week is labelled as such.
Attribution
Schedule, forecasts, previous values and impact classifications: Forex Factory. Dates and figures may be revised. Confirm important events with the source.

Observation is different from interpretation

An observation is a value published by a named source for a stated reference period. Pulse calculates comparisons using defined rules, then explains the evidence in plain language. Interpretations include supporting observations, counter-evidence and limitations. They are not instructions to place an order.

Codex prepares structured explanations from a fixed evidence packet. Numerical facts are inserted from validated observations rather than maintained independently in prose. Automated checks reject missing references, inconsistent definitions and invalid arithmetic. A well-formed explanation can still be wrong; those checks do not certify economic reasoning.

The market scanner adds a separate, transparent alignment score. Higher policy rates, stronger growth or a recent exchange-rate change do not by themselves establish the next exchange-rate move. The rules and limits are below.

How the scanner scores work

The scanner compares technical, fundamental and positioning evidence. Scores run from −10 to +10. They are research heuristics, not a calibrated winning probability, a return estimate or an instruction to trade.

Balanced: technical 40%, fundamental 40%, positioning 20%. The positioning budget is split between leveraged-fund COT (60% of that budget) and the non-reportable sentiment proxy (40%). These are four visible components from three data families. The two CFTC components do not receive two independent full weights.

Alternative lenses: Trend first uses 60% technical, 20% fundamental and 20% positioning. Macro first uses 20%, 60% and 20%. These are local exploratory views; the published default stays Balanced.

A total at or above +2 is labelled Bullish; at or below −2 is Bearish; otherwise Mixed. Labels use the unrounded total; the displayed score is rounded to one decimal. The scales and neutral bands are disclosed choices, not empirically established predictors.

Technical

Three equal votes compare price with MA20, MA20 with MA50, and MA50 with its level five observations earlier. Each difference has a ±0.1% neutral band. Average the votes and multiply by ten. Moving averages use only past observations. RSI14 uses Wilder smoothing and is context, not another vote.

Fundamental

Policy-rate difference contributes 50% of the factor, aligned quarterly growth momentum 30%, and aligned three-month labour momentum 20%. The policy difference is divided by two percentage points and capped. The US range midpoint is labelled. Growth and labour compare within-economy changes over common periods before comparing their directional votes. Growth uses a ±0.1 percentage-point neutral band and labour ±0.05 points; each pair component is half the difference between the two economies’ votes. Inflation is shown as context and is not automatically scored positively or negatively.

COT and sentiment proxy

Leveraged-fund net positions are divided by open interest, then scaled so a net share of ±20% reaches the component’s cap. The sentiment proxy is the non-reportable long-minus-short balance divided by long-plus-short positions. It is a futures-positioning category, not broker retail data or a survey of beliefs.

EUR/USD uses EUR futures against USD. USD/JPY inverts JPY futures. EUR/JPY subtracts the yen’s normalised tilt from the euro’s and divides by two, not an invented direct report for that cross. USD is a quote reference, not an observed neutral position. Both legs must use the same position date.

Coverage and freshness

A required missing or stale family makes the total unavailable; it is never silently treated as neutral. Freshness is checked when a snapshot is prepared and again in the interactive scanner. Static views preserve their labelled snapshot date. Reference prices expire for scoring after seven calendar days, COT after fifteen days from the position date, and the macro briefing after seven days. Each source keeps its own observation date. No historical performance claim is made.

Rules version: pulse-fx-alignment-v1. The interface and every score are computed from the same frozen inputs and versioned rules.

European Central Bank

ECB daily reference-rate history

Retrieved by Pulse
2026-09-20 06:08:18.083 UTC
Publication cadence
Daily reference observations on ECB publication days.
Pulse calculations
Same-date currency crosses, reference-rate returns, moving averages and RSI. These are not intraday or New York closing prices.
Limits
Price-based signals depend on the ECB reference convention. No open, high, low, volume or execution price is inferred.
Reuse
Reuse with source acknowledgement and identification of changes. Technical indicators and derived crosses are calculated by Pulse.

Commodity Futures Trading Commission

CFTC TFF futures-only positions

Retrieved by Pulse
2026-09-20 06:08:22.392 UTC
Publication cadence
Weekly positions normally refer to Tuesday and are normally released Friday. Position date and retrieval time are shown separately; no exact release time is inferred from the position date.
Pulse calculations
Leveraged-fund net contracts divided by open interest; non-reportable long/short balance; pair orientation and relative cross-currency tilts. Both positioning components share one capped weight budget.
Limits
Futures positioning is not the spot FX market. Non-reportable positions are a sentiment proxy, not broker retail percentages. The USD is the quote reference for the selected futures, not an inferred neutral USD position.
Reuse
CFTC government information is public domain; the CFTC requests acknowledgement. Only the selected public aggregate fields are used.

Comparisons need matching definitions

Annual inflation compares an index with the same month one year earlier. A change between two inflation rates is expressed in percentage points. Missing months are not filled or treated as zero.

The US GDP headline is an annualised quarterly rate. Where a pair comparison converts it to an ordinary quarterly rate, the calculation is ((1 + annualised rate ÷ 100)¼ − 1) × 100. The original rate remains visible. Seasonal and calendar adjustments are disclosed; national measures are not identical definitions of economic conditions.

Pulse compares actual observations with prior observations. It does not have a consensus-expectations dataset and does not label a release as exceeding or falling short of market expectations.

Ten-market factor table

The factor table covers a selected set of FX, metal, crypto and index market families. Broker aliases identify the family; they do not imply identical contract specifications or quotes. FX technicals use ECB reference observations. CFTC columns use the explicitly named futures contract, not a direct measure of the whole spot or CFD market.

Blue and red cells show positive and negative alignment with the named rule. Grey zero is neutral; a dash is unavailable. Inflation is contextual and has no directional vote. MA20/MA50 and slope columns explain the technical factor and are not counted again. A total is unavailable when required price, macro or positioning coverage is incomplete.

Gold uses managed-money positions in the disaggregated futures-only report. The other mapped markets use leveraged funds in the financial futures-only report. These are different classifications. Non-reportable positioning is a futures crowd proxy, not broker retail sentiment. DAX has no contract mapping in this integration.

The three US equity indices share an economic backdrop, so they must not be treated as three independent macro opportunities. Likewise, FX carry rules do not become a gold or equity valuation model. For GBP/USD labour comparisons, ONS rolling three-month periods are labelled by their end month and compared with an average of the same three US monthly unemployment readings. The national estimation methods still differ. Australian quarterly GDP growth is calculated from consecutive seasonally adjusted chain-volume levels. No seasonal or predictive-performance claim is made.

Factor table evidence

ECB reference exchange rates

Use and attribution
Source acknowledgement and identification of transformations. Derived same-date crosses are calculated by Pulse.
Successful retrieval
2026-09-20 06:08:24.150 UTC
Observation timing
Daily ECB reference observations, not executable quotes.

Factor table evidence

CFTC financial futures positions

Use and attribution
Government public information, with CFTC acknowledgement.
Successful retrieval
2026-09-20 06:08:26.096 UTC
Observation timing
Weekly futures positions. Position dates and retrieval times are distinct; exact release timestamps are not inferred.

Factor table evidence

CFTC disaggregated gold futures positions

Use and attribution
Government public information, with CFTC acknowledgement.
Successful retrieval
2026-09-20 06:08:26.512 UTC
Observation timing
Weekly futures positions. Position dates and retrieval times are distinct; exact release timestamps are not inferred.

Factor table evidence

BIS UK and Australian policy rates

Use and attribution
BIS permitted use with acknowledgement; policy observations may lag announcements.
Successful retrieval
2026-09-20 06:08:27.134 UTC
Observation timing
Daily policy observations with dissemination lag. Missing holidays are not filled.

Factor table evidence

ONS UK quarterly GDP

Use and attribution
Source: Office for National Statistics licensed under the OGL v3.0.
Successful retrieval
2026-09-20 06:08:28.625 UTC
Observation timing
Monthly or quarterly official statistics; periods and statistical conventions are preserved.

Factor table evidence

ONS UK unemployment

Use and attribution
Source: Office for National Statistics licensed under the OGL v3.0. Three-month windows are preserved.
Successful retrieval
2026-09-20 06:08:28.700 UTC
Observation timing
Monthly or quarterly official statistics; periods and statistical conventions are preserved.

Factor table evidence

ONS UK consumer inflation

Use and attribution
Source: Office for National Statistics licensed under the OGL v3.0.
Successful retrieval
2026-09-20 06:08:28.771 UTC
Observation timing
Monthly or quarterly official statistics; periods and statistical conventions are preserved.

Factor table evidence

ABS Australian quarterly GDP

Use and attribution
Based on Australian Bureau of Statistics data. CC BY 4.0. Pulse derives quarterly growth from seasonally adjusted chain-volume levels.
Successful retrieval
2026-09-20 06:08:30.041 UTC
Observation timing
Monthly or quarterly official statistics; periods and statistical conventions are preserved.

Factor table evidence

ABS Australian unemployment

Use and attribution
Source: Australian Bureau of Statistics. CC BY 4.0; seasonally adjusted total unemployment rate.
Successful retrieval
2026-09-20 06:08:32.091 UTC
Observation timing
Monthly or quarterly official statistics; periods and statistical conventions are preserved.

Factor table evidence

ABS Australian consumer inflation

Use and attribution
Source: Australian Bureau of Statistics. CC BY 4.0; all-groups annual inflation, unadjusted.
Successful retrieval
2026-09-20 06:08:33.740 UTC
Observation timing
Monthly or quarterly official statistics; periods and statistical conventions are preserved.

Factor table evidence

Gold API gold quote

Use and attribution
Commercial website use is expressly permitted by the provider. Pulse displays a timestamped captured quote; it is not a history feed and contributes no technical score.
Successful retrieval
2026-09-20 06:08:40.611 UTC
Observation timing
Captured provider quote with its original UTC update timestamp. This is context, not price history or an input to the technical score.

Factor table evidence

Gold API bitcoin quote

Use and attribution
Commercial website use is expressly permitted by the provider. Pulse displays a timestamped captured quote; it is not a history feed and contributes no technical score.
Successful retrieval
2026-09-20 06:08:40.853 UTC
Observation timing
Captured provider quote with its original UTC update timestamp. This is context, not price history or an input to the technical score.

Interactive market feeds

The market explorer uses TradingView’s official Advanced Chart and Technical Analysis embeds with their own data. The selected symbol and provider are shown beside the chart. Forex, gold and most index views use OANDA; Bitcoin uses Bitstamp, and Germany 40 uses FOREX.com. Index feeds are related CFDs, not official cash-index fixings. Prices can differ from your broker.

The embed loads on request and controls its own data timing and availability. Technical summaries are shown where the selected feed supplies them. A feed without verified summary data displays its price chart without an empty summary panel. Indicators remain available through the chart controls. The provider’s technical readings and price data are not extracted into Pulse’s scores or snapshots. An embedded chart may move while the published factor table remains unchanged.

TradingView widget documentation ↗

Three dates, three meanings

Reference period is when the measured activity occurred, or when the reference rate was observed. Release time is when the provider published a value, where that time is known. Retrieval time is when Pulse successfully obtained the source. Updating a page does not make an old observation new.

Some feeds do not provide an exact original release time for each observation. Pulse leaves it unknown rather than substituting the retrieval time or a calendar estimate. A later revision may change a historical value, even when the reference period is unchanged.

Monthly and quarterly observations naturally remain unchanged between releases. Known release calendars and conservative age checks help identify delays. A failed retrieval keeps the last validated observation and its original retrieval time. Readers should check the briefing date and observation periods.

The selected calendar includes US policy, CPI, employment and GDP releases; ECB policy meetings; Eurostat inflation, unemployment and GDP releases; and Japan nationwide CPI, labour and GDP releases. It is not exhaustive. Where the official calendar supplies only a date or explicitly marks an event as all-day, Pulse does not invent an announcement time. Known release times are converted to the browser’s timezone with daylight-saving rules.

A record that does not acquire hindsight

Each briefing uses one versioned evidence snapshot. Later data revisions create new observations for future briefings. They do not silently replace the evidence in an earlier edition.

Corrections link the amended briefing to the original and explain the change. Initial editions establish a baseline; Pulse does not create backdated briefings to imply an earlier publication history. Browse the briefing archive.

Charts and attribution

Interactive market charts use TradingView Lightweight Charts™. Copyright (c) 2026 TradingView, Inc. The library is distributed under the Apache License, Version 2.0; its original attribution notice is provided unchanged.

The library draws data supplied by Pulse; it does not provide the exchange-rate feed. The reference-rate chart contains daily ECB observations and disclosed cross-rate calculations. It does not manufacture open, high, low, close or volume observations. Accessible tables remain available alongside charts.

Expanded FX research desk

Daily references across the currency universe

The pair explorer uses ECB daily reference rates. The core set contains the 28 combinations of EUR, USD, GBP, JPY, CHF, CAD, AUD and NZD. Extended pairs compare EUR and USD with the other currencies in the latest source response. These are reference-rate combinations; their inclusion does not promise that a broker offers the same instrument.

Source: European Central Bank. Cross rates, changes and indicators are calculated by PineConnector Pulse. Latest reference date: 18 Sept 2026. Successfully retrieved 2026-09-20 06:08:42.664 UTC. The source does not supply an exact release timestamp for each historical observation.

Every source rate is quoted as currency units per euro. For A/B, Pulse divides B-per-EUR by A-per-EUR on the same date. EUR has a value of one for that calculation. Missing observations are not filled or joined to another date. The historical series is the source’s current vintage and can include revisions; it is not a record of what was known at each historical publication time.

The screener compares the same endpoints across all pairs for its 1, 5, 20 and 60 reference intervals. Missing endpoints remain unavailable. These intervals exclude non-publication days and are not exact calendar weeks or months. Chart labels 1M, 3M, 6M, 1Y and 2Y select up to 20, 60, 120, 260 and 519 reference intervals; the actual dates remain visible. Comparisons use common dates and start each series at zero percent.

Trend, volatility and range

Rising, Falling and Mixed describe the same three moving-average comparisons used by the scanner’s technical factor: reference versus MA20, MA20 versus MA50, and MA50 versus its value five observations earlier. Each difference has a ±0.1% neutral band; the three votes are equally weighted. This adds price research, not macro or positioning coverage, to the expanded pairs.

Volatility is the sample standard deviation of 20 daily log reference returns, multiplied by √252 and expressed as an annualised percentage. This convention is not a forecast. An internal missing observation makes that daily-return window unavailable. Range position compares the latest reference with the lowest and highest of the last 20 reference levels. A flat range is unavailable. These observations cannot measure intraday highs, lows, spreads or executable risk.

Observations older than seven calendar days display a stale notice. Historical data remains available with its original date. Failed retrievals do not replace the last validated snapshot or advance its successful-retrieval time. None of these tools produces an entry instruction or a calibrated prediction.

Currency strength and the performance map

Strength is the geometric mean percentage change of each of the eight core currencies against its seven peers. Pulse averages the log changes, reversing their sign when the currency is the quoted leg, then converts the mean back to a percentage. All 28 pairs use common reference dates. Every peer must be available; none is silently omitted or replaced by zero. The fixed basket does not change when the screener is filtered. Displayed percentage strengths need not sum to zero.

The currency map shows the row currency’s change against the column currency. Its horizon uses the same global reference-date endpoints as the screener. Both directions are calculated directly, so inverse returns are not assumed to be equal and opposite. A missing endpoint is unavailable. Opening a cell displays the pair’s standard quote direction, which may reverse the map cell.

Return correlation and comparison

The correlation matrix calculates Pearson correlation across 20 or 60 matched daily log reference returns. Both the start date and the end date of each return must match. Fewer than the requested observations, or a flat return series, produces an unavailable result. The displayed coefficient ranges from −1 to +1.

Positive correlation describes returns moving together; negative correlation describes opposite movement. Shared currencies can create mechanical overlap. These historical relationships can change and do not establish causation, future diversification or independent confirmation of a view. Selecting a cell opens both pairs on a chart rebased to their common first observation.

Saved pairs are stored in the browser on this device. No sign-in or data provider connection is required to use the desk.