Rebuild the financial data layer for the AI-native investor.
The infrastructure that powers institutional analysis — accurate, point-in-time SEC fundamentals across 30+ years — costs five figures a year and lives behind black-box terminals. Valuein opens that layer up: the same primary source, the same depth of coverage, queryable from any AI agent, Python notebook, or browser. Cheap enough for an individual analyst. Honest enough for a backtest. Open enough to compose into whatever you're building next.
105M+
Standardized Financial Facts
19,000+
Companies Covered
1993
History Starts
8
Parquet Tables Per Tier
Why this exists
The financial data market is broken in three quiet ways. It is too expensive for individual analysts and small funds — Bloomberg starts at $24K/year per seat, WRDS academic licenses are six figures, and every “cheap” alternative tops out somewhere short of usable.
It carries survivorship bias — datasets that drop bankrupt and delisted companies make every backtest look 200 bps better than reality. Strategies that only work because Enron isn't in your training data aren't strategies.
And it is not point-in-time — most vendors silently overwrite amended filings, so when you query “Apple's 2018 revenue” you get the restated value, not what the market actually saw on January 1, 2019. That single bug invalidates most published quant research.
Valuein fixes all three. The same SEC EDGAR primary source everyone else uses, parsed and standardized into a queryable Parquet warehouse, with accepted_at on every fact and every delisted ticker preserved. Free for S&P500 history, $49/mo for the full universe.
How Valuein runs
Everything is automated and edge-native. The data pipeline runs on schedule, ingests the latest SEC submissions, normalizes XBRL into ~200 canonical concepts, runs validation checks, and pushes Parquet files to per-tier object storage. The MCP server, Bulk Data API, and Python SDK all read from the same warehouse — no duplication, no drift.
The frontend, the API, and the rate limiter all run at the edge. There is no kubernetes cluster to babysit, no separate ops team, no hand-off between “the people who built it” and “the people who maintain it.” The same authors who ship the pipeline answer the support email.
That model has limits — Valuein isn't the place for white-glove account management or custom data delivery contracts. What it offers instead is honest documentation, a real status page, public methodology, and a release cadence you can audit in the changelog. If those trades fit the way you work, the free tier is one click away.
Two channels. One dataset.
Whether you're a Python quant or building an AI-powered workflow, Valuein meets you where you work.
- Python SDK →
pip install valuein-sdk (or uv pip install valuein-sdk) — DuckDB-backed DataFrames in 60 seconds. SQL-native, no black-box abstractions, no API key required to start.
- MCP Server →
One Bearer token, every MCP-compatible client. Claude, Cursor, Codex, ChatGPT — all query SEC fundamentals as a first-class tool.
Try it before you buy it.
Sample tier is one click — no signup, no credit card. S&P500 companies, 5 years of history, free forever.