Data catalog
Reference pages for the datasets, MCP tools and endpoints Hindcast is building. Each one describes what is covered and how it stays point-in-time. Endpoints open to the waitlist first, and the votes decide the build order.
55 pages · 4 categories
Datasets
The eleven catalog datasets, one page each — sources, fields and how they are kept point-in-time.
SEC & EDGAR data for AI agents (API + MCP)
SEC filings, XBRL fundamentals, 13F and Form 4 — normalized, as-reported and citable, behind one MCP server and REST API for AI agents.
Brazil market & fund data API (CVM · B3 · BCB)
Brazilian market data for AI agents: CVM fund portfolios and daily reports, B3 prices, BCB series, Tesouro, IBGE and CNPJ registry through one API.
Central bank data API — statements, minutes, speeches
Fed, BCB, ECB, BoE, BoJ and LatAm central banks: statements, minutes, speeches and projection vintages, timestamped and citable for AI agents.
Point-in-time macro data API (vintage series)
Vintage macroeconomic series for backtests and agents: query GDP, CPI and employment as they were first published, not as later revised.
13F & fund holdings API (US + Brazil, point-in-time)
Institutional holdings for AI agents: 13F filings and Brazilian CVM fund portfolios, resolved to issuers and kept point-in-time by disclosure date.
Market data for AI agents — EOD prices & deep history
End-of-day equities, ETF and futures history for agents and backtests: corporate actions applied, delisted names retained, survivorship handled.
Global media transcripts dataset (earnings, news, podcasts)
Earnings calls, world news, radio and podcast transcripts — searchable, speaker-attributed and timestamped, with citations for every extract.
Entity resolution API (LEI ↔ CIK ↔ CNPJ ↔ ISIN ↔ ticker)
One identity graph across jurisdictions: map LEI, CIK, CNPJ, ISIN, CUSIP and tickers to a single entity, with history for renames and reorganizations.
CFTC COT & positioning data API for agents
Commitments of Traders and positioning series normalized weekly, aligned to release dates so agents and backtests never read a report early.
Prediction market data API (Polymarket · Kalshi)
Polymarket and Kalshi market history normalized and point-in-time: contract terms, resolution criteria and price history an agent can cite.
Commodities & agro data API (EIA · USDA · Brazil)
Energy and agricultural data for agents: EIA series, USDA and WASDE releases, and Brazilian agro data, each aligned to its publication date.
MCP servers
What an MCP-capable agent can ask for, by topic. One remote server, tools per dataset.
MCP server for SEC filings — EDGAR tools for agents
Give your agent EDGAR as tools: search filings, pull XBRL facts and cite accession numbers over MCP, without writing a scraper.
MCP server for CVM fund data (Brazilian funds)
Query Brazilian fund portfolios and daily reports from your agent over MCP: CVM CDA positions and informe diário, resolved by CNPJ.
MCP server for macroeconomic data (vintage-aware)
Let your agent query macro series as of any past date over MCP: first prints, revisions and release calendars instead of latest-value-only.
MCP server for 13F holdings — institutional positions
Institutional holdings as agent tools: query 13F positions, quarter-over-quarter diffs and issuer resolution over MCP, filed-date aware.
MCP server for earnings call transcripts
Search earnings calls from your agent over MCP: passage-level retrieval with speaker attribution and timestamps, cited back to the call.
MCP server for central bank speeches & communication
Central bank speeches as agent tools: search by speaker, bank and date over MCP, with venue and delivery timestamps on every passage.
MCP server for FOMC statements and decisions
FOMC statements as agent tools: retrieve any meeting's statement over MCP and diff it against the previous one, with release timestamps.
MCP server for CFTC COT reports (positioning)
Commitments of Traders as agent tools: weekly positioning by category over MCP, with release timestamps so nothing is read early.
MCP server for Brazilian market data (B3 · BCB · CVM)
One MCP endpoint for Brazil: B3 prices, BCB series, CVM funds and CNPJ resolution, so an agent can research the market without local plumbing.
MCP server for fund portfolios (cross-border holdings)
Fund portfolios as agent tools: US 13F positions and Brazilian CVM portfolios in one schema over MCP, joined by resolved issuer identity.
MCP server for the BCB Focus survey (market expectations)
Brazil's Focus market expectations survey as agent tools: query any weekly vintage over MCP to see what forecasters expected at the time.
MCP server for treasury data (yields & reference rates)
Government bond yields and reference rates as agent tools: US and Brazilian curves over MCP, aligned to publication dates.
MCP server for commodity data (EIA · USDA · Brazil agro)
Commodity fundamentals as agent tools: EIA energy series, USDA and WASDE vintages and Brazilian agro data over MCP, by release date.
Financial data MCP server — one endpoint for market data
One remote MCP server for financial market data: filings, fundamentals, macro vintages, holdings, transcripts and prices, with citations.
Use cases & endpoints
Specific queries and endpoints people search for, from XBRL facts to COT reports.
Point-in-time fundamentals API — as-reported financials
Query company fundamentals as they were originally reported, with revision history intact, so backtests and agents never see a later restatement.
As-reported financial data (originals, not restatements)
Get the numbers companies actually filed, not the tidied later versions: as-reported financials with restatements kept as separate observations.
XBRL financials API — normalized company facts
XBRL company facts without the taxonomy pain: normalized concepts, resolved units and filing provenance on every value, for agents and pipelines.
Form 4 insider transactions API for agents
Insider buys and sells from SEC Form 4: transaction and filing dates, reporting-owner roles and resolved issuers, structured for agents.
FOMC dot plot history — every projection vintage
Every FOMC dot plot vintage, queryable by release: see how the distribution of rate projections shifted between rounds, with sources attached.
Fed SEP projections data — growth, inflation, rates
Summary of Economic Projections as structured data: growth, unemployment, inflation and rate projections by round, with central tendencies and ranges.
FOMC minutes API — full text, searchable and dated
FOMC minutes as structured text: searchable by section and meeting, with both the meeting date and the release date on every document.
Central bank speeches dataset — searchable and cited
A structured corpus of central bank speeches: speaker, role, venue and delivery date, chunked for retrieval and linked to the source document.
BCB Focus survey API in English (market expectations)
Brazil's Focus survey with English field names and every weekly vintage: expectations for inflation, Selic, growth and FX, by horizon and release.
Copom minutes in English — Brazil's rate decisions
Brazil's Copom statements and minutes with English alongside the Portuguese original, dated by release so as-of questions stay honest.
CVM CDA fund portfolio data (Brazilian holdings)
Brazilian fund portfolios from CVM CDA archives: positions and weights by CNPJ and competência, with disclosure dates for point-in-time queries.
CVM informe diário API — daily fund NAV and flows
Daily Brazilian fund data from CVM informe diário: NAV, net flows and shareholder counts by CNPJ, as a clean series instead of monthly archives.
CNPJ to ticker mapping — Brazilian company identity
Map a Brazilian CNPJ to its listed tickers, ISINs and global identifiers, with dated mappings that survive renames and restructurings.
LEI, CIK and CNPJ mapping API for cross-border data
Cross-border identifier mapping: resolve LEI, CIK and CNPJ to one entity so US filings and Brazilian registry data join without manual reconciliation.
ISIN to ticker resolution API (instruments, dated)
Resolve ISINs to tickers and venues with dated mappings, so historical price and holdings joins do not attach to a recycled symbol.
CFTC COT report API — weekly positioning data
Commitments of Traders as a clean weekly API: positions by trader category, legacy and disaggregated formats, stamped with release timestamps.
Kalshi historical data — event contract price history
Historical event contract data: prices, volumes, resolution terms and settled outcomes, timestamped so past probabilities stay as they were quoted.
Earnings call transcripts API — searchable with citations
Earnings call transcripts as structured data: prepared remarks and Q&A separated, speakers attributed, passages timestamped and citable.
WASDE report data API — supply and demand estimates
USDA WASDE estimates as structured data with every monthly vintage retained, so the market-moving revision between reports is directly computable.
EIA energy data for AI agents (inventories & production)
EIA energy series for agents: inventories, production and consumption aligned to release dates, with revisions kept as separate observations.
Brazil agro data API — production, exports, harvest
Brazilian agricultural data for agents: production estimates, export flows and harvest progress aligned to publication dates and producing regions.
Survivorship-bias-free data — delisted names retained
Historical universes that include the companies that failed: delisted securities retained with final prices, so backtests are not run on survivors.
Financial data for LangChain agents (tools + citations)
Wire market data into LangChain agents: filings, fundamentals, macro vintages and holdings as tools, each returning a citable source.
Financial data for LlamaIndex — retrieval over filings
Retrieval-ready financial text: filings, transcripts and central bank documents chunked with provenance, for LlamaIndex pipelines and agents.
Financial data for CrewAI — shared tools across agents
Give a CrewAI crew one shared market data layer: consistent identifiers and cited values, so agents in a workflow do not contradict each other.
Financial data for Claude via MCP — connect and query
Connect Claude to market data over MCP: filings, fundamentals, macro vintages and holdings as tools, with citations Claude can quote back.
Guides
Why point-in-time correctness matters, explained with real examples and primary sources.
Look-ahead bias in backtesting — how it happens
How look-ahead bias creeps into backtests through revised data, filing lags and survivorship, and the data properties that prevent it.
As-reported vs restated data — which one should you use?
The difference between as-reported and restated financials, when each is correct, and why storing only one of them breaks historical analysis.
What is point-in-time data? (bitemporal, explained)
Point-in-time data stores when a value refers to and when it was published, so you can query what was knowable on any past date. Here is how it works.
GDP first print vs revisions — why the number changes
US GDP is published in a sequence of estimates and revised for years afterwards. Here is why the first print differs and when to use each vintage.
Join the waitlist. Vote the roadmap. First in, first served. Join the waitlist.