Financial Data API for Macro, Positioning and Fed Sentiment

Macro data for AI agentsThe data layer for financial intelligence.<br>Financial Data API is one API and one MCP server for macro history, futures positioning and Fed sentiment. Every number is stored as it was known at the time and traces back to an official source.

Get your free API keyRead the docs<br>No card required. 2,500 calls a month free, key issued in under a minute.

Point in time data

Revisions are kept, so backtests never see the future

Positioning extremes

COT percentiles for every CFTC contract since 1992

Fed sentiment scored

Every FOMC statement since 1994, from dovish to hawkish

Official sources only

FRED, ECB, OECD, BIS, SEC, CFTC, EIA and the central banks

Derived signalsWhat can you pull from the API?<br>Derived signals with receipts. Every value is stored, versioned and traceable to the official release behind it.

COT positioning percentilesGold printed 93.3 the week of its 2011 top and 2.9 at the 2015 bottom. Every CFTC contract, weekly since 1992, with crowded long and crowded short flags.Point in time valuationAAPL traded at 13.7 times trailing earnings in the middle of 2012. All 13 multiples run daily since 2007, built from prices as traded and fundamentals as reported on the day.FOMC statement sentimentEvery Fed statement since 1994 gets a score from a fixed, versioned lexicon. December 2008 reads fully dovish. June 2022 reads fully hawkish.Screen macro like stocksFilter countries by macro conditions in a single request. Inflation, unemployment and policy rates across the covered universe.Ask it from your agentA hosted MCP server takes one URL plus your key. Claude, Cursor and other agents can then discover and query every dataset with no glue code.<br>Gold futuresNet positioning percentile<br>CFTC · weekly<br>93.3 · 2011 top2.9 · 2015 low2010 – 2015, every week storedCrowded long

Built for backtestsHow deep is the history?<br>COT to 1986, housing to 1959, OECD confidence to 1960 and energy to the early 1980s. Full published history, not snapshots.

Positioning depth

Weekly COT for every CFTC contract, stored back to 1986 with percentiles since 1992.

Breadth of record

Official sources across macro, rates, FX, energy and fundamentals, some to 1959.

Positioning history972 CFTC contracts<br>Connected<br>Official sources87 connected<br>Connected<br>FOMC statements244 scored<br>Connected

The Fed, on the record

Every FOMC statement since 1994, scored on a fixed and versioned lexicon.

Agent nativeWhich AI agents can use it?<br>A hosted MCP server, llms.txt and an OpenAPI spec let an agent discover and call every endpoint with no glue code.

Connect your agent

Research gradeWhy do quants trust the data?<br>Built for quants, research desks and regulated teams that audit what they use.

Ready for backtests<br>Query any date with as_of and get the data as it was known then. Revised data never leaks into the past.

Clean rights<br>The public API serves only official data that is safe to redistribute. No licensing surprises.

Full provenance<br>Trace any number to its official source, the source URL, the raw payload reference and the ingestion run.

Deep history<br>COT to 1986, housing to 1959, OECD confidence to 1960 and valuation multiples to 2007.

Easy to call<br>Cursor pagination, stable error codes, a typed SDK and a hosted MCP server.

Secure by default<br>Scoped API keys, rate limits per plan and HTTPS across every surface.

The proofNumbers you can check yourself<br>Three worked examples, live on the API today. Pull them with your own key.

Gold futuresNet positioning percentile

93.3week of the 2011 topCrowded long<br>2.9week of the 2015 bottomWashed out

Was the trade crowded?<br>Gold printed a 93.3rd percentile net long the week of its 2011 top and 2.9 at the 2015 bottom. Every CFTC contract, every week, back to 1992.

See the COT data

AAPL trailing price to earningsDaily, point in time

13.7× · June 2012<br>2007today

Multiples you can backtest<br>AAPL traded at 13.7 times trailing earnings in the middle of 2012. All 13 trailing multiples run daily since 2007, built from prices as traded and fundamentals as reported on the day.

Explore valuation

FOMC statement scoreJune 2022 reading

DovishHawkish

Hawk or dove, in one number<br>Every FOMC statement since 1994 gets a score from a fixed, versioned lexicon. December 2008 reads fully dovish. June 2022 reads fully hawkish.

Read the methodology

Use casesWhat do people build with it?<br>The same observation model serves research, monitoring, agents and product features.

Backtest macro and positioning strategies on data that never leaks revisions into the past.<br>Systematic researchQuant desks and individual quants

Give a trading agent live macro context through the hosted MCP server. One URL and a key.<br>AI agentsClaude, Cursor and custom agents

Flag crowded trades with COT percentiles. Every CFTC contract is scored weekly against its own history.<br>Positioning monitorsFX, rates and commodity desks

Chart 15 years of valuation multiples...