Actionable insights — Who is Chris Hohn?

1. Screen for tollkeepers: a toll that survives a change in what passes over it

1. Describe the business as a toll: what is the thing that must pass through it, and what fraction of the value does it keep?
2. Confirm the three properties — high barriers to entry, structural pricing power, predictable cash flows. All three, not two.
3. Run the substitution test: change the technology of what passes over the toll and see whether the toll survives. Electric cars still need roads; a Mastercard transaction still crosses a payment network.
4. Prefer tolls whose demand is compelled — by regulation, by physical necessity, or by an installed base — over tolls that depend on the customer choosing you each period.
5. Sanity-check with duration: a genuine toll should show a very long revenue record, not a strong decade.

• A toll whose rate is politically exposed — regulated tolls can be capped as easily as they are protected (interchange fees, ratings-agency reform).
• Confusing a strong brand with a toll; a brand can be walked past, a toll cannot.

2. Require five or six overlapping barriers, and name every one

1. List the candidate's moats explicitly against a fixed menu: intellectual property, brand, hard assets, long-term contracts, network effects, regulatory or switching costs.
2. Count them. One barrier is a thesis with a single point of failure; the target is five or six that each independently deter an entrant.
3. For each barrier, ask what a well-funded competitor would have to do to neutralise it, and how many years that would take.
4. Reject names where the count is padded — two descriptions of the same advantage are one barrier.

• Barriers that all rest on one underlying fact (a single regulator, a single customer contract) — correlated moats fail together.
• An eroding barrier being replaced silently by marketing spend; the count stays the same while the quality falls.

3. Convert pricing power into a number: real price rise ÷ margin

1. Estimate how much above inflation the company can raise price each year without losing volume — from history, not from hope.
2. Take the operating margin.
3. Because a price rise costs nothing to deliver, it lands almost entirely in profit: profit growth exceeds revenue growth by roughly the real price rise divided by the margin.
4. Use the result to rank candidates. A 1% real price rise on a 20% margin adds ~5 percentage points to profit growth; on a 50% margin it adds ~2.
5. Then ask the qualitative question the formula assumes: would customers actually absorb the rise, and what stops them shopping around?

• Price rises that are really mix shift; the same formula applied to a mix improvement will overstate durability.
• Volume decay that shows up two or three years after the price rise, once contracts renew.

4. Run a risk-first mandate — and let it exclude things outright

1. Replace the opening question. Instead of "how much can I make," ask "how much can I lose," and answer it before any valuation work.
2. Apply a comprehension gate: if you cannot explain the balance sheet or the earnings mechanism in plain language, the name is excluded — not sized down, excluded.
3. Test the gate on the people running it. If management cannot explain it either, treat that as decisive evidence rather than as a red flag to monitor.
4. Accept that the gate will exclude whole sectors, and write down which ones, so the exclusion is a policy rather than a mood.
5. Frame the mandate for yourself in one phrase you can be held to — Hohn's is "a stay rich fund, not a get rich fund."

• A comprehension gate that quietly relaxes during a bull market; the excluded sector is usually the one performing best.
• Complexity that has been outsourced rather than removed (off-balance-sheet vehicles, reinsurance sidecars, joint ventures).

5. Pair concentration with a long holding period — they are the same decision

1. Decide how many names you can genuinely underwrite to the standard above. If the barrier count is five or six per name, that number is small — Hohn's is 10-15.
2. Let the top five carry the majority of the book, and check the actual weights against that intention rather than assuming.
3. Set the holding period to match the analysis: work that takes a year to do should not be re-decided quarterly. Hohn's average hold is eight years, some over thirteen.
4. Justify the length with the persistence premise — "good companies stay good and bad companies stay bad" — and treat a violated premise, not a price move, as the sell trigger.
5. Accept the consequence: with a top position at a quarter of the book, one name will dominate any given year's result, up or down.

• Concentration achieved by drift rather than by decision — a winner growing to a quarter of the book is a position you have not re-underwritten at that size.
• The eight-year average being an artefact of a few very old holdings while newer ones churn.

6. Read another investor's book for the filter, not for the names

1. Take a disclosed portfolio and work backwards to the rule that admits every holding and excludes the obvious alternatives.
2. State the rule in one sentence. If you cannot, the portfolio is not evidence of a method.
3. Test the rule on names the manager does not own — a rule that only ever ratifies existing holdings is a description, not a filter.
4. Adopt the rule where it overlaps your own criteria and note honestly where it does not (Hohn's Microsoft position sits in a corner Slegers deliberately avoids).

• 13F-style disclosures that omit non-US or non-equity exposure — the post itself notes Safran and Airbus are excluded from the weights shown.
• Copying a concentrated book without the underwriting behind it; the weights are the output of the work, not a substitute for it.