The goal of this phase is not to ask the AI to predict next quarter’s EPS or generate a generic “buy/sell” rating. Long-term investors need a structured, comparative judgment on business quality, financial durability, capital allocation, and valuation.
This article provides junior investors and analysts with a prompt-driven workflow to extract that comparative judgment. By forcing Gemini to cross-reference management’s rhetoric with your uploaded 10-year financial tables, you can cut through the noise and identify true long-term compounders.
The Four Pillars of Durability
When analyzing stocks for a five- to ten-year holding period, every prompt you feed Gemini should serve one of four core questions:
- Business Quality: Does this company have a demonstrable moat?
- Financial Durability: Do margins, returns on equity (ROE), and free cash flow (FCF) show resilience across full economic cycles?
- Capital Allocation: Has management used cash in ways that drive long-term compounding, or destroyed it through poor M&A?
- Valuation Sanity: Are you paying a reasonable price for this quality relative to the company’s history and its peers?
To keep your analysis standardized across different themes, it helps to use established checklists. The Long-Term Stock Checklist provides excellent raw material that can be easily translated into Gemini ratio tables and prompt criteria. Additionally, utilizing written, step-by-step stock analysis frameworks ensures you aren’t skipping critical variables when designing your AI workflow.
Key takeaway: Ask Gemini for structured tables and cited evidence, not for investment advice. The AI surfaces the reality; you own the final decision.
1. Business Quality: Extract and Compare Moats
Your first analytical prompt should force Gemini to pull moat-related evidence from the uploaded documents. Crucially, the AI must separate management claims from financial evidence. A company that boasts about “network effects” in its investor deck but shows collapsing retention metrics in its financial tables does not have a strong moat.
For a deeper dive into how competitive advantages interact with financial outcomes, this breakdown of moat, numbers, and valuation frameworks is a highly recommended watch before you start prompting.
Prompt 1: Moat Comparison “For each company in this notebook, extract evidence related to competitive advantages from the uploaded annual reports and investor-day materials. Categorize claims under: switching costs, network effects, cost advantage, intangible brand/IP, and regulatory barriers. For each category, provide a short quote or paraphrase with the document and year cited. Finally, rank the companies by ‘evidence strength’ (strong / moderate / weak / unclear), specifically pointing out any disagreements between management’s qualitative claims and the 10-year ROE/margin tables uploaded earlier.”
Pro-Tip Follow-ups:
- “Which specific customer stickiness metrics appear in the filings (retention, churn, renewal rates, NRR)? Compare how each company defines them.”
- “List any regulatory licenses or standards mentioned that act as barriers to entry, noting any highlighted expiration dates or political risks.”
2. Financial Durability: Build 10-Year Trend Tables
Next, ask Gemini to assemble comparable tables using your uploaded metric summaries. You aren’t just looking for absolute numbers; you are looking for the shape of the data over time.
Prompt 2: Durability Tables “Using only the uploaded long-term metrics summaries and filings, build a year-by-year table covering the last 10 fiscal years for each ticker. Include: ROE, Operating Margin, Debt-to-Equity, and FCF per Share. Below the table, write a short paragraph per company classifying its financial profile as either: consistent & improving, consistent but flat, cyclical but resilient, volatile, or deteriorating. Highlight specific years that break the trend and pull quotes from that year’s filing explaining what caused the break.”
Interpret with care:
- A high ROE funded by aggressively rising leverage is a red flag, not a sign of operating excellence.
- Dividend growth that outpaces Free Cash Flow is a warning sign of an unsustainable payout.
- Margin expansion driven by slashing R&D or Capex will eventually reverse.
3. Capital Allocation: Follow the Cash
Compounders are made or broken by how management allocates capital. You want Gemini to map exactly where the cash went over the last decade.
Prompt 3: Capital Allocation Breakdown “From the cash flow statements and management discussions across the uploaded annual reports, summarize how each company allocated cash over the last 10 years. Categorize by: reinvestment (Capex/R&D), M&A, share buybacks, dividends, and debt paydown. Estimate rough percentage proportions. Assess whether this pattern supports long-term compounding: prioritize evidence of high-return organic reinvestment and disciplined M&A. Flag any serial large acquisitions, buybacks funded by new debt, or dividends that required leverage increases.”
You are looking for thesis drift. If your thesis relies on a capital-light software compounder, but Gemini reveals the company has quietly morphed into a debt-heavy acquisition machine, your thesis is broken.
4. Valuation Sanity: Simple Metrics, Explicit Assumptions
Long-term analysis requires a price check, but LLMs are notoriously bad at unprompted financial math. They can easily mishandle share counts, one-off tax benefits, and complex accounting quirks.
Keep the valuation metrics simple (P/E, PEG, Price-to-Book) and always provide your own growth assumptions rather than letting the AI hallucinate a consensus estimate.
Prompt 4: Valuation Sanity Narrative “Using the price series and earnings figures provided in the metrics summaries, compute the current P/E for each ticker and compare it to that company’s 10-year median. Compare the peers against each other on this same basis. Write a valuation sanity narrative: Which names look expensive versus their own history? Which look reasonable? Do the peer valuation gaps accurately reflect the quality differences we saw in the ROE/FCF durability tables? Do not recommend buy or sell. Explicitly state the math and assumptions (e.g., trailing vs. forward EPS) you used for these calculations.”
Always spot-check Gemini’s arithmetic against your source tables to ensure accuracy.
The Comparison Workflow Checklist
To run a flawless comparison session in Gemini Notebook, follow this sequence:
- Verify Context: Confirm the notebook contains filings, your thesis notes, and the 10-year metric summaries for every ticker.
- Run Prompt 1 (Moats): Extract competitive advantages and demand strict citations.
- Run Prompt 2 (Durability): Generate financial trend tables, classify the profiles, and investigate any “break” years.
- Run Prompt 3 (Capital Allocation): Map the cash flow to identify compounding-friendly vs. value-destructive management behaviors.
- Run Prompt 4 (Valuation): Execute a historical and peer-based valuation sanity check using your explicit assumptions.
- Synthesize: Take the outputs and write your own final concluding paragraph for each stock, noting what requires further human investigation.
Analyzing for long-term durability is a comparative craft. Gemini accelerates the extraction and tabulation of data, but it does not replace your judgment. The investors who benefit most from AI treat the model as a tireless, brilliant junior analyst—one that is excellent at lining up the evidence, but never allowed to own the final conclusion.