TL;DR: Run the same checks in the same order on every candidate, before the story gets exciting: what the business does and how it makes money, whether the financials confirm it, what the price already assumes, what could prove you wrong, and the price or event that would make you sell. A checklist does not make you right. It makes your mistakes consistent enough to learn from.
Why a checklist, and why in a fixed order?
Because the failure mode it prevents is specific. Investors rarely lose money on a company they researched and got wrong. They lose it on the company they half-researched: the earnings call they meant to finish, the debt figure they never looked up, the valuation they skipped because the story felt obvious. Pilots and surgeons use checklists for the same reason. Not because the steps are hard, but because attention is unreliable exactly when confidence is high.
The order matters too. Business first, valuation last. If you open with the price, every later judgment bends toward justifying it. If you open with the business, the price arrives as the final question it should be: given all of the above, what am I being asked to pay?
Step 1: The business, in plain sentences
- What does it sell, and who pays? If you cannot say it in two sentences without jargon, you do not understand it yet. Stop here until you can.
- Why do customers stay? Switching costs, habit, network effects, brand, contracts, or nothing in particular. "Nothing in particular" is an answer; it just prices the business differently.
- Who else wants this profit pool? Name the two or three strongest competitors and what would have to happen for them to take share. Competition you cannot name is competition you have not looked for.
- What does management say it is doing, and did last year's version happen? Compare this year's stated priorities with what the same documents promised two years ago. Follow-through is visible in the record, not in the charisma.
Step 2: The financials, as a story over five years
One year of numbers is a photograph. Five years is a plot. Pull the trend, not the snapshot:
- Revenue growth: steady, accelerating, decelerating, or cyclical? Each pattern needs a different kind of faith.
- Margins: expanding, stable, or quietly eroding while revenue grows? Growth bought with margin usually sends the bill later.
- Cash conversion: does reported profit become free cash flow, or does it accumulate as receivables and inventory? Profit that never turns into cash is a claim, not a result.
- The balance sheet: how much debt, against how much cash generation, maturing when? Debt decides who survives a bad year, and bad years are not optional.
- Share count: falling through buybacks at sensible prices, flat, or creeping up through dilution? Per-share numbers are the only ones you own.
The margins in that story come from one document above all, which is why how to read an income statement sits inside this step.
Step 3: The price, and what it assumes
Only now does valuation enter, and it enters as three separate tools because each answers a different question. A ceiling check like the Graham Number asks whether the price respects today's earnings and assets. A reverse DCF asks what growth the price already assumes, so you can compare the market's expectation with the history you just read in Step 2. A forward model asks what the business is worth if your own assumptions hold.
The checklist rule is simple: write down what has to be true for this price to work, in numbers. "It has to grow cash flow about 12 percent a year for a decade" is a research conclusion you can test later. "It seems reasonably valued" is not.
Step 4: The risks, written as falsifiable statements
Generic risk lists ("competition, regulation, macro") teach nothing. Write risks that could actually be checked:
- Not "competition is a risk," but "if the largest rival's new product takes the mid-market tier, revenue growth halves; I will see it first in gross margin."
- Not "management risk," but "the growth story depends on the acquisition program; two of the last three deals were written down."
- For each risk: what would you observe, where would you observe it, and how soon?
Then the question most checklists skip: what is the strongest case against this purchase, stated by someone who believes it? If you cannot write that paragraph, you have read only one side's research, which usually means the selling side's.
Step 5: The exit, decided in advance
Before buying, write down:
- The price or result that proves the thesis wrong. Not a stop-loss percentage; the evidence. "Two consecutive years of negative free cash flow" is evidence. "Down 20 percent" is weather.
- The review date. A fixed date, a quarter or two out, when you re-run this checklist against fresh numbers regardless of how the position feels.
- The position size the completed work has earned. A half-finished checklist has not earned a position at any size; finish the steps first. Size is how you price your own uncertainty, and unfinished research is the largest uncertainty on the page.
This step feels pessimistic on the day you are most optimistic, which is exactly why it has to be written that day. The version of you reading it in six months will be very glad you did.
How does the checklist fit a screening workflow?
The checklist is the second pass, not the first. Screening narrows thousands of stocks to a handful worth these hours; the tools page collects the calculators this checklist calls, and a screen built on fundamentals, growth, margins, dividends, and balance-sheet strength produces candidates in the shape the checklist expects. Then the checklist slows you down on the few finalists, in order, on paper.
And when a finalist passes all five steps, one discipline remains free and underused: paper trade the position first while you watch it through one earnings cycle. The checklist grades the company. Paper trading grades you.
Educational content, not financial advice. A checklist organizes research; it does not predict outcomes, and every figure above must be verified against the company's own filings before any decision.
