ferminrp/dataroma-superinvestors

DataRoma Superinvestors via Markdown

Consult DataRoma through defuddle markdown pages and return clean, agent-friendly summaries.

Core URLs

• Index of active managers: https://defuddle.md/https://dataroma.com/m/home.php
• Holdings by manager code: https://defuddle.md/https://dataroma.com/m/holdings.php?m={managerCode}
• Activity (all): https://defuddle.md/https://dataroma.com/m/m_activity.php?m={managerCode}&typ=a
• Activity (buys): https://defuddle.md/https://dataroma.com/m/m_activity.php?m={managerCode}&typ=b
• Activity (sells): https://defuddle.md/https://dataroma.com/m/m_activity.php?m={managerCode}&typ=s

For URL templates, parsing expectations, and examples, read references/endpoints.md.

Workflow

1. Fetch manager index from home.php via defuddle.
2. Resolve the requested manager and extract the m= code from the holdings link.
3. Fetch holdings for that manager.
4. Fetch activity with typ=a, and optionally typ=b or typ=s if user asks specifically.
5. Return a concise markdown report prioritizing holdings quality and recent moves.

Output Contract

Use this structure by default:

1. Manager Snapshot
   • Manager name
   • Period
   • Portfolio date
   • Number of stocks
   • Portfolio value
2. Top Holdings
   • 8-15 rows prioritized by % of Portfolio
   • Columns: Ticker | Company | % Portfolio | Recent Activity | Shares | Reported Price | Value
3. Recent Movements
   • Group notable rows by Add, Reduce, Buy, Sell
   • Include share change and % change to portfolio when available
4. Quick Take
   • 2-5 bullets with actionable interpretation (concentration, largest adds/reduces, unusual moves)
5. Sources
   • Include DataRoma original URLs and the defuddle URL used

Data Quality and Fallback Rules

• If manager name is ambiguous or not found, offer 3-5 close candidates from the index and ask user to choose.
• If activity markdown is noisy or incomplete, prioritize holdings and state the limitation explicitly.
• Preserve original DataRoma links for traceability.
• Treat this as informational market data; avoid financial advice language.
• If a field is missing, show N/A instead of inferring values.