Actionable insights — 13 Interesting Holding Companies

1. Hunt for discounts caused by analytical cost, not by distress

1. Ask why a security is cheap, and sort the answers into two piles: something is wrong with the business, or something is expensive about understanding it.
2. Prefer the second. A discount that exists because valuing the thing takes a week of work is available to anyone willing to spend the week; a discount that exists because earnings are falling is not.
3. Test the diagnosis by counting how many separate businesses you would have to value to reach a number. The more there are, the more durable the discount is likely to be.
4. Do the sum of the parts anyway, and record it, so that the next time the price moves you only have to update it.

• A discount that never closes. Complexity keeps buyers away permanently, so the return has to come from the underlying assets compounding, not from the gap narrowing.
• Your own intrinsic-value number drifting with the price. Note that $68 is used against a $42 price here and against $46.5 in the portfolio update four days later — the same value estimate, two different discounts.

2. When the asset is a capital allocator, judge the allocator — not the operations

1. Establish first that the company genuinely allocates rather than operates: it owns pieces of businesses and earns from appreciation and dividends, not from selling goods.
2. Then evaluate the only variable that matters — the decision-making record. Length of tenure, the family or steward in control, and the size of their own stake.
3. Score the structure rather than the underlying industries. Two holding companies with identical assets can compound at 8% and 19% depending on who is choosing.
4. Ignore revenue and margins entirely for this category; they describe subsidiaries you may not even own outright.

• Family control treated as automatically good. It also means minority holders cannot force a change when the allocation record turns — EXO.AS at 1.5% a year since 2022 is the case in point.
• An allocator whose record is mostly one lucky position rather than a repeated process.

3. Open every holding company by its top five weights, before anything else

1. Pull the disclosed top-five positions and their percentages first; they usually decide most of the outcome.
2. Add up how much sits in the single largest name, and ask whether you would own that name directly at that weight.
3. Treat any large cash line as a position too — it is a bet on the manager's future decisions, not on the current portfolio.
4. Only then read the narrative about strategy and themes.

• A thematic label that the weights do not support. Two named themes and one 44% position is a single-stock bet with a theme attached.
• Cash disclosed as a position. It is honest reporting and it is also 27% of your money doing nothing until the family finds something.

4. Split the portfolio into what has a market price and what has an estimate

1. For each holding company, calculate the share of net asset value that is quoted daily versus the share carried at a manager's or auditor's estimate.
2. Treat the estimated share as the part of your investment you cannot verify, and size accordingly.
3. Look for a named future event that would convert an estimate into a price — a listing, a partial sale, a third-party round.
4. Where realised exits are disclosed, compare the exit price against the previous carrying value to measure how conservative the marks have been.

• A daily-priced wrapper around an illiquid book. The listing gives you liquidity; it does not give the underlying assets a price.
• Private marks that only ever move upward between funding rounds.

5. Count the fee layers between you and the assets

1. Map the chain of ownership from your share to the operating asset, and identify every entity that charges a fee along the way.
2. A listed vehicle investing in third-party funds has two layers; an investment trust has one; a true operating holding company has none.
3. Deduct the layers from the historic return before comparing across candidates — or check whether the quoted CAGR is already net.
4. Where possible, take the exposure through the entity that collects the fee rather than the one that pays it.

• Historic CAGRs quoted gross in one case and net in another, then compared side by side.
• A holding company that has quietly become a fund manager — which changes what you are buying.

6. Distinguish a discount case from an access case, because they need different tests

1. Ask why the wrapper exists for you: to buy assets below their worth, or to buy assets you could not otherwise reach at all.
2. For a discount case, the test is the sum of the parts against the price.
3. For an access case, the test is different — whether the private assets are worth owning at their carried value, and whether the wrapper's fee is a fair toll for the access.
4. Never let one case borrow the other's justification; an access vehicle trading at a premium is a coherent thing to own, and a discount vehicle is not made better by having a famous private position in it.

• An access vehicle whose access is already priced in. The premium the market pays for the SpaceX exposure is the actual entry cost, and it is not stated here.
• Access arguments used where the assets could in fact be bought another way.

7. Build the category list with its failures included, and read the dispersion

1. When compiling a list of a structural category, include the poor performers with their numbers printed at the same prominence as the good ones.
2. Read the dispersion rather than the average: it tells you how much of the outcome comes from the structure and how much from the specific manager.
3. Where the spread is wide, treat the category as a hunting ground, not as an allocation — you must pick within it.
4. Then require a name-specific reason for every candidate you take from the list.

• Start dates chosen per name. "Since 2001", "since the 2006 IPO", "since 1993" and "since its listing in 2022" are not comparable, and Brookfield is quoted two different ways within a single post.
• Lists with no price on any entry. As with the Lindy series, these are universe-building exercises — treat them as a source of candidates, and do the valuation yourself.