0:00 1. Disqualify fast — "say no in 60 seconds"
The repeatable method
- Treat idea triage as elimination, not selection: with ~60,000 listed stocks the goal is to reject, not to find reasons to buy.
- Give each candidate a 60-second test: look for a single disqualifier (no moat, no founder/skin-in-the-game, capital-hungry, weak cash conversion). "In 98% of cases you can find a reason to say no."
- Sort what survives into three piles (Munger): yes, no, and "too hard / too complex to understand" — and be willing to leave the too-hard pile alone indefinitely.
Here: he applies it live — his community asks "what about company X?" and he finds the no within a minute; only a handful (Kinsale, Medpace, Dino Polska) clear every filter and get real work.
Watch for
- Any one criterion failing on a first pass — that's a "no," and it saves the deep-dive for the rare survivors.
4:57 2. Compete only where you have an edge — the behavioural one
The repeatable method
- Classify any edge you might have as informational, analytical, or behavioural.
- Concede the first two: information is all online now, and analytical firepower (quants, PhDs, AI) crushes the individual, especially on large, widely-followed caps.
- Press the one edge institutions can't use: behaviour. You have no clients to answer to, so you can underperform for a quarter, hold cash, and wait for the obvious pitch — patience and discipline are the whole advantage.
Here: he owns zero big tech precisely because that's the crowded, analytically-efficient corner where retail has no edge — and still beat the S&P (+40% vs +24% since Oct 2023).
Watch for
- An idea whose thesis depends on knowing/modelling something better than Wall Street on a mega-cap — a red flag you're playing where you can't win.
13:20 3. Screen every name against the same six quality criteria
The repeatable method
- Require all six: (1) a moat; (2) skin in the game — a founder still running it with the majority of his wealth in the stock; (3) low capital intensity; (4) good capital allocation; (5) high profitability; (6) attractive growth.
- Treat skin-in-the-game as quantitative, not vibes: an HBR study found founder-led firms outperform ~3.9%/yr — enough, compounded, to put you in the top 1–2% of investors on its own.
- Run the whole market through the filter to build a fixed "investable universe" (his came to ~156 names); only shop from that list.
Here: MEDP founder August Troendle holds $1.9B of a $2B net worth in the stock; KNSL and DNP.WA are founder-run — each clears all six.
Watch for
- Insider/founder ownership %, tenure, and whether the founder is still operating; any of the six missing = out.
15:42 4. Test earnings quality with FCF conversion
The repeatable method
- Set a profitability floor (net margin ≥10%), then check the more important number: what fraction of earnings turns into free cash flow.
- Require ≥90% earnings-to-FCF conversion. "Earnings are an opinion; cash flow is a fact." Poor conversion flags possible financial engineering, weak ethics, or simply a bad business.
- Use it as a quality tiebreak: the top FCF-conversion decile has historically beaten the bottom decile by ~18% (Jeremy Siegel, Stocks for the Long Run).
Here: the criterion underpins his cash-first read of every pick, and it's the axis on which he separates a compounder from an accounting-earnings story.
Watch for
- Net income that persistently outruns free cash flow (conversion <90%) — a signal to dig into working capital, capitalised costs, and accruals.
17:28 5. Set growth thresholds, then haircut the forecasts
The repeatable method
- Demand >7% revenue growth and >9% FCF-per-share growth, both realised (past 5–10 yrs) and expected.
- Don't trust analyst forecasts at face value — they're systematically too optimistic; apply a Graham-style margin of safety, cutting 30–40% (a "15%" forecast becomes ~11% in your model).
- Prefer management guidance over analyst estimates (firms underpromise/overdeliver), and let a long-tenured CEO with a strong track record stand as evidence of durable execution.
Here (20:33): he explicitly discounts a 15% analyst estimate to ~11% before it drives a valuation.
Watch for
- A thesis that only works on un-haircut sell-side growth; whether the end market itself is structurally growing (payments, obesity, pet care) so the growth is tailwind-driven, not heroic.
22:59 6. Judge capital allocation by ROIC — but only with a reinvestment runway
The repeatable method
- Assume most CEOs are poor capital allocators (they rose as salespeople/operators, untrained for the job) — so make allocation the thing you spend the most time on.
- Measure it with return on invested capital (NOPAT ÷ invested capital); require >15%. At 20% ROIC, every $100 invested throws off $20 of value.
- Apply the Munger caveat: a high ROIC only compounds if the company can reinvest most of its cash at that rate. A high-ROIC business that can't reinvest (so it buys back stock) is good; one that reinvests everything at a high ROIC is the rare "golden goose."
Here: DNP.WA (Dino Polska) is the golden goose — reinvesting nearly all FCF at a high return; AAPL is the cautionary limit — great ROIC, but too much cash to reinvest, hence buybacks.
Watch for
- ROIC vs the reinvestment rate: is the company able to deploy retained cash back into the business at a similar return, or is it forced to return it?
40:17 7. Triangulate value with three quick methods
The repeatable method
- Forward PE vs its own 10-yr average — naive, ignores the current outlook, but a stock at its cheapest multiple in a decade is a flag worth chasing.
- Earnings-growth model — expected annual return ≈ EPS growth + dividend yield ± change in the multiple. Decide if that number clears your hurdle.
- Reverse DCF — instead of forecasting, solve for the FCF growth the current price already implies, then judge whether that's beatable versus history/guidance. (Invert, per Munger.)
Here: LVMUY flags on method 1 (cheapest PE in 10 yrs); KNSL ≈ 13% EPS − 1% multiple + 0% yield ≈ 12%/yr on method 2; MEDP's price implies only ~11% FCF growth vs ~15% historical on method 3 → "too conservative," a buy.
Watch for
- Agreement across all three; a reverse-DCF implied growth rate that sits comfortably below the company's demonstrated growth = margin of safety baked into the price.
44:48 8. Value on a napkin — roughly right beats exactly wrong
The repeatable method
- Don't chase two-decimal precision on intrinsic value; the inputs are too uncertain to justify it.
- If a business is genuinely cheap and good, it should be obvious from a back-of-a-napkin calculation.
- Use complexity as a signal: "if you need Excel to figure out whether a stock is interesting, it probably isn't."
Here: his Kinsale return math (a three-term sum) and Medpace reverse-DCF read are both stated off the top of his head — decisions that don't survive without a model are treated as non-decisions.
Watch for
- A thesis whose attractiveness hinges on fine-tuned assumptions — a sign the margin of safety is too thin.
46:25 9. Sell only when the case breaks — never on valuation
The repeatable method
- Default to buy-and-hold: if the investment case is intact, the company keeps compounding intrinsic value, so there's no reason to sell — don't trim winners on price.
- Sell for exactly one reason: the original thesis is no longer intact (the moat, the growth, or the predictability you underwrote has changed). Treat every sale as evidence you made a mistake at purchase.
- Stress-test with the closed-market test: "would you still own this if the stock market shut for 10 years?" Yes → keep; no → sell.
Here: TXT.WA sold when AI flipped from tailwind to threat (predictability gone);
ULTA sold when the moat proved thinner (beauty retail more competitive/fragmented) — while his biggest, now-expensive winner stays untrimmed.
(49:41)
Watch for
- A change in the business (technology threat, competitive erosion), not the price — that's the only valid sell trigger. Rising valuation alone is not.