Research any stock with Claude: the one-prompt scorecard

What you need

• 🧠 A Claude account at claude.ai (web search is available on every Claude plan and cites its sources: claude.com/blog/web-search)
• 🎨 Artifacts turned on. They are Claude's interactive mini-apps and work on the free plan too: what are artifacts
• 🏛️ Nothing else. The filings come from the SEC's EDGAR database, which is free public access for everyone: about EDGAR

Step 1: pick your ticker

• 🎯 Any US-based public company works. They must file a 10-K annual report (audited financials) plus quarterly 10-Qs: what a 10-K is. Foreign companies listed in the US file a 20-F annual report instead, and the prompt still works, Claude just pulls that.
• 🔎 Want to see the raw filings yourself first? Search any company at EDGAR full-text search

Step 2: paste the master prompt

Swap in your ticker and paste the whole thing into a new Claude chat:

You are an equity research analyst building a pre-screening risk scorecard. The company is [TICKER].

Step 1: research. Use web search and cite every figure as you go.
- Pull the company's latest 10-K (annual report) and most recent 10-Q filings from SEC EDGAR.
- Pull the last four quarters of revenue, gross margin, operating margin, EPS, and free cash flow.
- Pull current valuation context: market cap, trailing and forward P/E, EV/EBITDA, price/sales, price/book, PEG.
- Pull balance sheet basics: cash position, total debt, debt/equity, current ratio.
- Pull growth signals: revenue growth, segment growth, margin trajectory.
- Pull the 3 biggest upcoming catalysts and the 3 biggest risk factors (start from the 10-K risk factors section, then recent news).

Step 2: score. Rate three pillars from 0 to 100 and show your working.
- Valuation: cheap vs stretched, relative to growth and to the company's own history.
- Financial health: balance sheet strength, liquidity, margins, cash generation.
- Growth: revenue and segment trajectory, margin direction.
Weight them 40/30/30 into one overall score. Bands: 0-39 high risk, 40-69 mixed, 70-100 healthier profile. Give every metric a one-word tag (e.g. stretched, conservative, improving).

Step 3: build. Create an interactive artifact (single-page React, dark theme) with:
- A header card: ticker, company name, price context, the overall score on a gauge, and a three-sentence verdict in everyday language.
- A valuation grid, a financial health grid, and a growth grid, one card per metric with its tag.
- A quarterly trend table for the last four quarters.
- A catalysts vs risks two-column panel.
- A bottom line section written like an analyst note, three sentences max.
- A sources tab listing every figure with the URL it came from, grouped by section.

Rules: if you cannot verify a number, mark it "unverified" instead of guessing. Prefer primary sources (SEC filings and the company's investor relations page) over blogs. Add a footer that says this is a research aid, not investment advice.

Step 3: read it like an analyst, not a fan

• 📊 Start with the three pillar scores, not the overall number. The shape tells you more than the total.
• 🚩 Any red-tagged metric is a question to answer, not a verdict, click into why it got that tag.
• 📅 The quarterly trend table is where stories fall apart or hold up. One great quarter is noise, four in a row is a pattern.

Step 4: verify the three numbers that drive the score

• 🔗 Open the sources tab and click through on the two or three figures doing the most work in the verdict.
• 🏛️ Cross-check them against the actual filing on EDGAR: EDGAR full-text search
• ✏️ If something is off, tell Claude "recheck this figure against the 10-Q and update the scorecard", it will correct and re-render.

Step 5: make it reusable

• ♻️ Save the master prompt as instructions in a Claude Project, then a new scorecard is just typing a ticker.
• 🗓️ Re-run it after each quarterly 10-Q drops so the trend table stays current.
• ⚖️ Compare two companies by running the same prompt twice, the identical rubric is what makes the comparison fair.

Honest limits (read this bit)

Why this works so well now

• 🏦 The filings Claude reads are the same public documents professionals use. EDGAR serves them free to everyone: about EDGAR
• 💼 Anthropic even builds a dedicated financial services offering on top of Claude for institutions doing due diligence and research: Claude for financial services
• 💸 The data terminal this replaces for a hobby-level workflow runs about $31,980 a year per seat in 2026: Bloomberg Terminal pricing