Actionable insights — Our Shopping List (Part II)

1. Choose the valuation yardstick from the business model, and say why

1. Ask what the company's earnings actually represent before choosing a multiple. Lumpy, mark-driven or float-derived earnings do not support a P/E.
2. For an insurer or investment holding company, use book value. For a fund-like wrapper, use NAV. For a serial acquirer with heavy amortisation, use cash earnings.
3. State the chosen multiple and convert it to a price, so the yardstick is auditable.
4. Then stick to it. Changing basis between notes on the same company destroys the ability to check yourself.

• The same company later priced on a different basis. Fairfax gets an intrinsic value of CAD 3,000 on 30 April and is bought at CAD 2,300 in August with no valuation at all — 29% above the target set here.
• "Cheap on one measure, expensive on another" inside a single write-up: this one calls Fairfax "a great business at a cheap price… just 8x earnings" and simultaneously 28% overpriced on book.

2. Price a candidate against the position it would sit beside, not against the market

1. Identify the closest existing holding — same industry, same economics, ideally the direct duopoly partner.
2. Put the two multiples side by side on the same basis.
3. Ask what the difference buys you: is the premium justified by growth, quality or diversification, or is it just a different ticker?
4. If the answer is "nothing much", the candidate is a duplication of risk rather than an addition.

• Adding the second name in a duopoly and calling it diversification. It is the same regulatory, network and interchange-fee risk twice.
• Benchmarks chosen after the fact to make a candidate look cheap.

3. Treat a persistent margin gap against direct peers as the strongest available evidence of a moat

1. Find two or three genuine direct competitors — same product, same customers.
2. Compare net margin, on the same definition, over several years rather than one.
3. If the gap persists through a full cycle, look for the mechanical cause: automation, scale, mix, or pricing.
4. Cross-check with a headcount or asset-per-employee figure, which is harder to manipulate than a margin.
5. Ask what would close the gap, and whether anyone has the incentive to try.

• Margin gaps that come from a different revenue mix (interest income in a high-rate period) rather than from lower costs.
• Your own familiarity bias: this is the archive's own broker, named in every issue's sources footer.

4. Price a cyclical against its own historic multiple range, not against a quality peer group

1. Plot the company's forward multiple across at least one full cycle and note the extremes.
2. Establish where in the cycle the current earnings sit — for a housing supplier, against housing starts.
3. Set the entry multiple deliberately between the extremes, and accept it may not be reached for years.
4. Never pay a peak multiple on peak volumes; that is the double-count that ruins cyclical returns.

• Structural arguments (a housing shortage, energy-efficiency regulation) being used to justify paying a cycle-peak multiple.
• EPS growth of "over 1,000% since 2016" measured from a cycle trough — the start date is doing a lot of work.

5. Separate moats that rest on approvals and records from those that rest on technology

1. Ask what a well-funded competitor would actually have to do to compete, step by step.
2. Count the steps that cannot be shortened with money — regulatory approvals, elapsed operating history, certified part counts, safety records.
3. Those are the durable ones. Technology gaps close; time does not compress.
4. Check whether the customer is allowed to take the risk of switching, not just whether they would want to.

• A regulatory moat that regulation can also remove. FICO loses 55% within two weeks of an agency approving a rival score.
• Paying any price for a moat. This one is passed on at 46.7x here and again at 53-57x in the July deep dive.

6. Look for businesses that benefit from someone else's price increases

1. Identify companies positioned as the cheaper alternative to a supplier with pricing power.
2. Check whether the discount is expressed as a percentage. If so, every price rise by the incumbent widens the absolute saving.
3. Confirm the customer cannot avoid the purchase — maintenance, compliance, insurance-mandated spend.
4. Then the incumbent's greed is your holding's tailwind, and it requires no action from management.

• The incumbent cutting price to defend share, which collapses the arbitrage.
• Fleet age reversing when deliveries normalise — the tailwind is a delivery-backlog artefact as much as a structural one.

7. Verify a scarcity strategy in the secondary market

1. For a brand that deliberately under-supplies, find the resale price of its core product.
2. Compare it to the retail price. The ratio measures unmet demand directly, which no survey or brand-value study can.
3. Check who controls supply: family or founder control is usually what allows scarcity to survive quarterly pressure.
4. Then judge whether the discipline is likely to hold — the risk to this model is management chasing volume, not customers losing interest.

• Resale premiums narrowing — the earliest sign that scarcity has been over-monetised.
• Store count creeping up. 300 stores worldwide is part of the mechanism, not an accident.

8. Audit your own published targets against what you actually paid

1. Keep every target price you publish, with its date and its basis.
2. When a name is eventually bought, compare the price paid to the target and note the difference.
3. If they diverge, say which of the two was wrong — the target was too demanding, or the discipline slipped.
4. Watch particularly for a change of valuation basis between the target and the purchase. That is how a target gets quietly retired.

• A moving intrinsic-value estimate doing the work a fixed multiple used to do.
• Fourteen names priced across Parts I and II, one inside its limit, and then three purchases that go to existing holdings instead. That is a legitimate outcome, but it means the shopping list did not drive the capital.