Actionable insights — Position Switch

1. Sell on relative opportunity within a category, and pre-allocate the proceeds

1. Group your holdings by business model, so like-for-like comparisons are possible.
2. Within a group, rank on valuation against expected growth — not on absolute cheapness.
3. When one name trades at the same multiple as its peers on materially worse growth, that is the funding source for the others.
4. Decide where the money goes before you sell, and deploy it in full so no cash drag is introduced.
5. Say plainly that the sale is a relative judgement, not a verdict on the business.

• The stated rule and the actual trade diverging. The post says the only valid reason to sell is a broken thesis, then sells for a better opportunity. Both sentences are in the same article.
• Concentration by model. Swapping one roll-up for two others increases exposure to the serial-acquirer model, not diversification.

2. Treat a large acquisition by a serial acquirer as a warning, not a milestone

1. Track deal sizes as a percentage of the acquirer's own market value, not in absolute currency.
2. Flag any deal that is a step change from the historical average — the model's safety comes from many small, uncorrelated bets.
3. Ask what single-asset risk the large deal reintroduces: cyclicality, project lumpiness, customer concentration.
4. Watch the following two years of segment results specifically for that asset.

• Management framing a large deal as transformational. That is the same fact, described the other way.
• The acquired asset's revenue recognition. Project-based businesses inside a recurring-revenue roll-up break the model's predictability.

3. Reconsider the case when a founder stops being chief executive

1. Record for each holding whether the founder-led status is part of why you own it.
2. When the founder steps down, re-run the case from scratch rather than assuming continuity.
3. Distinguish a departure into a chairmanship with a large stake from a full exit.
4. Set a review date rather than reacting immediately, and check capital-allocation decisions made afterwards.

• The reverse case. PRM on 28 May is bought largely because proven operators arrived; the same variable in both directions.
• Founders who leave with the stock at a low. The departure and the difficulty usually share a cause.

4. Limit exposure to companies whose customers are funded by a government budget

1. Trace the revenue back two steps: who pays your customer?
2. Where the answer is a public budget, treat the demand as a political variable, not an economic one.
3. Size the exposure as a share of revenue and check whether the budget line is currently being cut.
4. Discount the growth forecast accordingly, because analyst estimates lag political decisions.

• The same exposure elsewhere in the book. Any life-sciences tools or research-instrument name carries it.
• Budget cuts that reverse. Political variables are volatile in both directions, which cuts against selling at the bottom of one.

5. Price stagnation explicitly: judge the multiple against the growth rate, not its own history

1. Establish the actual growth achieved over the last several years, using a fixed start date.
2. Add the consensus forecast for the next several years to get the full stagnation window.
3. Ask what multiple a genuinely no-growth business deserves, and compare with what you are paying.
4. Do not anchor on the stock's own historical multiple, which was set when it was growing.

• Cyclical troughs mislabelled as stagnation. Research funding is cyclical and political; six flat years is evidence, not proof.
• Selling at 17.5x a business you still say will "continue to do well" — the multiple has to be judged against the reinvestment opportunity too.

6. Buy after forced selling ends, not while it is happening

1. Identify price falls caused by a mechanical seller — margin calls, index deletion, fund redemptions, lock-up expiry — rather than by the business.
2. Confirm the business is unaffected, so that the fall is genuinely technical.
3. Wait for evidence the forced selling is finished. Buying into an ongoing liquidation means catching a falling supply curve.
4. Check the second-order risk: a founder who was margined once may be again.

• The unnamed source. "I heard from a great source" is doing the decisive work here and cannot be verified from public filings.
• Founder leverage as a recurring feature rather than a one-off. It is a governance fact, not just a past event.

7. Value each business on its sector's own cash metric

1. For businesses where statutory earnings are distorted by acquisition accounting, find the measure the industry actually uses.
2. For software roll-ups, use free cash flow available to shareholders; for accounting consolidators, profit after tax before amortisation of acquired intangibles.
3. Express the result as a yield or a forward multiple ladder across several years, so the growth is visible in the valuation.
4. Compare the resulting number to the company's own history, since cross-company comparisons on bespoke metrics are unreliable.

• Metrics that exclude a real cost. Amortisation of acquired software is non-cash, but the cash was spent buying it.
• Multi-year multiple ladders. The 9.2x is a 2029 number and depends entirely on the growth arriving.

8. Execute a switch as one transaction with published quantities and limits

1. Announce the sale and both purchases together, so the trade is judged as one decision.
2. Split the proceeds by a stated rule — here, equally — rather than by feel.
3. Publish quantity and limit price for every leg so execution can be verified later.
4. Add to existing positions rather than introducing a new name, so the switch does not also add research risk.

• The gap between the framing and the facts. Judges is called a weed in the closing line and "a company I still like" three paragraphs earlier.
• Three limit orders in three currencies that may not all fill. A partial execution turns a neutral switch into a cash position.