Introducing DEHY MCP — the data layer, native to Claude & CursorLearn more →

About

Why we built DEHY.

The signals that move markets are public — they just arrive slow, scattered and unstructured. DEHY reads the official record the moment it files, structures it to the field, and scores it, so the whole desk sees the same ground truth within a minute.

The record, made legible.

A Form 4 lands as forty pages of XML. A 13F is a quarterly wall of CUSIPs. The information is there — it just isn't usable at the speed a decision needs.

So we built one layer that reconciles every filing at the source, computes a 0–100 conviction score in the same pass, and serves it to a feed, an API and an AI assistant — no secondary feed, no scraping, no guesswork about where a number came from.

We would rather earn an analyst's trust than overclaim. The score is a triage signal — what to read first — not a promise about price, and the methodology is published in full, including the honest backtest.

Built on the primary record

SEC EDGARCFTCFederal ReserveFREDUSASpending
60M+
structured rows
506K
filings ingested
8,500
institutional filers
<60s
filing → structured

Methodology

A triage signal, not a prediction.

The DEHY Score ranks what to read first — it is not a claim about where a price will go. It is versioned, documented, and published with its backtest, warts and all.

Come see the record, structured.