How to Analyze a 13F Filing With AI
Turn a raw SEC 13F into a smart-money teardown — new buys, full exits, conviction changes — with one copy-paste prompt. Plus the delay caveat the paid trackers bury. Model-agnostic (Claude, ChatGPT, Copilot, Gemini, Perplexity).
For: Retail investors · analysts · anyone tracking smart moneyTo analyze a 13F filing with AI, paste the raw information table from SEC EDGAR — and, critically, the same manager's prior-quarter filing — into a fresh chat, then ask for a structured teardown rather than a summary. A working prompt instructs the model to: (1) list new positions opened this quarter; (2) list full exits; (3) list adds and trims with the change as a percent of shares held; (4) rank the top ten holdings by percent of the disclosed portfolio to read conviction; (5) flag concentration; and (6) state the data's limitations back to you. The two instructions that make it useful are “quote the exact share counts and values” and “compare against the prior quarter I pasted.” Then apply the discount that matters most: 13F data is up to 45 days stale and shows only long U.S. equities — never treat it as a live trade signal. The same prompt works in ChatGPT, Claude, Copilot, Gemini, or Perplexity.
Everyone wants to see what Buffett, Burry, or Pershing Square bought last quarter. A dozen paid trackers exist to sell you that view — and most of them quietly skip the part that decides whether the data is worth anything. A 13F is a legally required snapshot of a large manager's long U.S. equity book, filed up to 45 days after the quarter closed. Read correctly, it's a rich idea generator. Read as a live signal, it's a trap. The good news: you don't need a subscription to read one well. You need the raw filing, last quarter's filing, and a prompt that thinks like an allocator instead of a headline writer.
The workflow, step by step
- Pull the raw 13F from EDGARSearch the manager on SEC EDGAR (free), open the latest 13F-HR, and copy the information table — holdings, share counts, and market values.
- Grab last quarter's 13F tooThe signal is the change, not the snapshot. Pull the prior quarter's filing from the same manager so the model can compute deltas.
- Paste both into a fresh chatClean context avoids cross-contamination. Paste this quarter's table, then last quarter's, then the prompt below — all in one message.
- Run the smart-money teardown promptIt forces new buys, exits, adds/trims by percent, top-ten conviction, concentration, and an explicit limitations section.
- Apply the delay discountUp to 45 days stale, long-only, U.S.-listed only. Treat every read as a lagged, incomplete snapshot — then do your own valuation work.
The exact prompt (copy-paste)
Model-agnostic. Swap the [BRACKETS] and paste both filings into the same chat.
You are a long/short equity analyst reading institutional 13F filings. Above this prompt I have pasted [MANAGER]'s 13F for [THIS QUARTER] and, below it, their 13F for [PRIOR QUARTER]. Produce a smart-money teardown: 1. NEW positions opened this quarter (ticker, shares, market value, % of the disclosed portfolio). 2. FULL exits since last quarter. 3. ADDS and TRIMS: change in shares vs prior quarter as a % of the prior position. 4. TOP 10 holdings by % of the disclosed portfolio — this reads conviction. 5. Concentration: what % is in the top 5 and top 10 names? 6. LIMITATIONS: restate what this data does NOT show (delay, longs only, no shorts/cash/non-US/intra-quarter trades). Use only the numbers I pasted. Quote exact share counts and values. If a field is missing, say "not disclosed" — do not estimate. Output as a clean bulleted brief.
Get all 7 workflows in one pack
This is one of seven paste-ready workflows in The AI Finance Workflows Pack — the exact prompts, pro moves, and honest guardrails finance pros run at their desk. Free.
Instant access on the next page + the free weekly brief. Unsubscribe anytime.
Prefer to browse first? Open the Workflows Pack → · Finance Prompt Library →
Frequently asked
Can AI analyze a 13F filing?
Yes. Paste the raw 13F information table (and ideally the prior quarter's) into a model and prompt it for a structured teardown — new buys, exits, adds/trims, position sizing as a percent of the disclosed portfolio, and concentration. The value is comparing two quarters to see the change. Paste the primary-source EDGAR data rather than asking the model to retrieve it, and verify every figure. Educational only, not financial advice.
How current is 13F data?
13F filings are due 45 days after quarter-end, so a filing can describe positions up to about four and a half months old by the time you read it. They disclose only long U.S.-listed equities, options, and convertibles — not shorts, cash, non-U.S. holdings, or intra-quarter trades. That is why a 13F is a research starting point, not a live signal. Do your own research (DYOR).
Should I copy a hedge fund's 13F positions?
No. Cloning 13F positions ignores the 45-day delay, the hedges and shorts you cannot see, the fund's cost basis and time horizon, and its risk tolerance versus yours. Use a 13F to generate ideas and spot conviction changes, then do your own valuation and risk work. This is not a recommendation to buy or sell any security.