Smart Finance Projects

The three rules

1. 01

   Primary sources only

   The numbers come from filings with the U.S. Securities and Exchange Commission — 10-K, 10-Q, 8-K — pulled from EDGAR and compared line by line against the document. The comparison itself is published in the repository.

2. 02

   Assumptions with a name on them

   No growth rate, margin or discount rate enters a model without a one-line written reason from me. While the reason is missing, the field shows up as pending instead of being quietly filled in.

3. 03

   A model you can move

   The discounted cash flow runs in your browser. Move growth, margin, WACC and terminal growth, and the implied price recalculates. You can copy a link that carries your assumptions.

Featured report

All reports

How a report gets made

Six steps, in this order. A report only gets published when the six are done.

1. 1

   Read first

   The 10-K (business, risk factors, management’s discussion) and the latest earnings call, read whole, with notes taken by hand before any summary is requested from an AI.

2. 2

   Rebuild the numbers

   A spreadsheet built from scratch, then reconciled against the XBRL data pulled from EDGAR. If a cell cannot be explained, it does not go in.

3. 3

   Write the assumptions

   Growth, margin, capex, working capital, tax rate, WACC and terminal growth — each with the reason behind it, in one line, signed by me.

4. 4

   Run the checks

   The engine flags a terminal value above 80 % of enterprise value, a terminal growth above the discount rate, projected margins above the historical record without a reason, and scenario probabilities that do not add up to one.

5. 5

   Write the report

   Ten blocks, from the one-page summary to “what would change my mind”. Then a written cross-examination: the AI attacks the thesis, I answer in writing.

6. 6

   Publish with a date

   Data date, version and changelog on the page. Updates every quarter, with the diff visible.

Full site methodology →

How AI is used here

Claude (AI) downloads and organises the filings, writes the calculation engine and its tests, and builds these pages. It does not write the thesis, the assumptions, the risks or the conclusions — those are my work, and each report carries its own disclosure saying exactly who did what.

What’s next

• Finish the Lululemon report: assumptions with their reasons, comparables, scenarios and the written thesis.
• A second report on Chipotle Mexican Grill (CMG) — the financial data is already downloaded from EDGAR.
• A 3–5 minute video per report, explaining the thesis without a script.
• Ten hard questions per report, published with their answers.
• Quarterly updates with a visible diff against the previous version.

Educational, not advice

These reports are my work as a student, published to learn in the open. They are not financial, investment or tax advice, and they are not a recommendation to buy or sell anything. Whether I hold a position in a company I cover is declared on the report itself, and every position I hold is published on the portfolio page. Verify every figure at its source before deciding anything.

Where my own positions are published →