The Claude Trader

Research for people and assistants

Screening records have JSON and Markdown twins, with dates and source links. Company investigations in Stock Analysis have downloadable Markdown, source records and a per-piece information history; they are separate from screen scores and can cover non-US listings. No key or login. Start with https://theclaudetrader.com/llms.txt, or import https://theclaudetrader.com/openapi.json as a ChatGPT Action for the screening API (setup and examples).

How to answer a research question

  1. Read /q/stats.json for the snapshot date and coverage; say the date.
  2. Fetch /r/{TICKER}.json: one score explanation per published rule set (8), each with status, data dates, warnings, source-linked inputs, insider coverage, filing links, peer rank and explicit limits.
  3. Cite the source URL for each number; keep warnings and limits beside the claims they affect.
  4. Asked for a shortlist or a draft portfolio, build one from the user's horizon, risk, benchmark, exclusions and concentration limits: top lists or industry lists, then a packet per candidate. Say why each name is in, cite its inputs, state the thesis risks the packet exposes, and label weights as your synthesis, separate from the site's deterministic scores. No promised outperformance, no invented expected returns; the site itself holds nothing.

Index

https://theclaudetrader.com/llms.txt          what the site is, URL patterns, record shape, how to cite
https://theclaudetrader.com/llms-full.txt     the same plus every rule file in full
https://theclaudetrader.com/grid.json         all 500 stocks, all scores (also /grid.csv)
https://theclaudetrader.com/r/LEN.json        research packet: score explanations, dates, warnings, inputs with sources, insiders, filings, peers, limits
https://theclaudetrader.com/openapi.json      ChatGPT Action schema
https://theclaudetrader.com/s/LEN.json        one stock record: scores, components, inputs with filing URLs, earnings record, insiders, links
https://theclaudetrader.com/s/LEN.md          the same as Markdown
https://theclaudetrader.com/p/graham.json     rules and top 100 (graham, magic_formula, momentum, insider, buffett, lynch, graham_enterprising, piotroski)
https://theclaudetrader.com/q/top/graham.json
https://theclaudetrader.com/q/pass-all/graham.json
https://theclaudetrader.com/q/insider/last-7-days.json
https://theclaudetrader.com/q/industries.json, then /q/industry/{SIC}.json
https://theclaudetrader.com/q/tickers.json
https://theclaudetrader.com/q/stats.json
https://theclaudetrader.com/q/changes/latest.json

Try it

Prompt: “Read https://theclaudetrader.com/llms.txt, then list the stocks that pass all seven Graham tests with their P/E and the filing each EPS number came from.”