Actionable insights — Portfolio Update 2026 (Part II)

1. Read a company's own buyback behaviour as its valuation opinion

1. Track repurchases quarter by quarter rather than by the announced programme size.
2. Note when buying accelerates into a fall — that is management saying the shares are cheap with its own balance sheet.
3. Note when it stops without the cash position changing. That is the more valuable signal, and almost nobody publishes it.
4. Treat a stopped buyback on a rising share price as a price ceiling set by the best-informed party.

• A programme authorised and not executed. KNSL's "$250 million buyback" is permission; check the shares actually retired.
• Buying that only offsets stock compensation. Share count, not dollars spent, is the measure.

2. Match the profit metric to the business model — then hold the multiple to that metric

1. Ask what accounting charge distorts this particular model: acquisition amortisation for a roll-up, consolidation for a holding company, growth capex for a scaling manufacturer.
2. Choose the measure the owner actually receives, and use it consistently for both the multiple and the reverse DCF.
3. Where you have no adjusted metric, do not pretend the raw multiple means the same thing.
4. Say which metric you used, so two companies at "31x" can be compared honestly.

• An adjusted metric compared with an unadjusted five-year average of itself. Both sides of the ratio have to be built the same way.
• Management's own plan used as the growth input without a haircut, as it is here.

3. Score risk separately from cheapness, and let it cap the rating

1. Complete the valuation work and record what the models say.
2. Then list the live risks — legal, regulatory, operational, criminal — as a separate item.
3. Allow the risk list to cap the rating one notch below what the numbers alone would support, and say so explicitly.
4. Do not net the two into a single adjusted number, because that hides which one moved.

• Risks that are permanent rather than temporary. Regulation and organised crime targeting the product are not a cycle.
• A shareholder yield funded from a shrinking earnings base — the yield falls with the earnings, so it is not a floor.

4. Being right about the business does not upgrade the rating

1. Separate three questions: was the drawdown temporary, was the business fine, and is the price attractive now.
2. Answer the first two from evidence — the results, and what insiders did with their own money.
3. Answer the third from the models, and let it set the rating regardless of how well the first two went.
4. Record the vindication anyway, because the point is the process, not the outcome.

• A vindication story used to justify adding. The recovery is the reason the price no longer qualifies.
• A long-term EPS estimate that collapses (2.5% here) while the reverse DCF demands 13.2%. That gap, not the multiple, is the real objection.

5. For a rollout, count openings; for a mature chain, count same-store sales

1. Establish where the business is in its expansion: filling the map, or finished.
2. While it is filling, make unit openings the headline metric and treat like-for-like as secondary.
3. Check the total revenue line to confirm the openings are actually covering the shortfall.
4. Write down the milestone — market saturation — after which the metric must switch back, so the change is planned rather than forced.

• The switch never being made. When the map is full, openings stop and like-for-like becomes the only metric there is.
• A record opening year that also raises the capital intensity. Check the openings are still self-funded.

6. Test a cost advantage by what happens in the soft part of the cycle

1. Identify the industry's pricing cycle and where it currently sits.
2. Ask whether the company grew through the weak leg, or only during the strong one.
3. Locate the structural source of the advantage in a published ratio — a combined ratio, a cost per unit, an operating margin against peers.
4. Treat the soft market as the opportunity: it removes competitors from a business that can survive it.

• Growth through a soft market bought with looser underwriting. The combined ratio is the check, and it lags.
• Reserve releases flattering the operating figure. Twenty per cent growth needs a source.

7. Audit the published sheet for your own errors, including the ones in your favour

1. Re-check every pass/fail mark against the numbers printed beside it, not against the conclusion.
2. Look specifically for marks that flatter a holding — those are the ones nobody queries.
3. Correct them in the record rather than in the next issue.
4. Where an entry disagrees with a sibling entry on identical evidence, say which one is the exception.

• Errors that all point one way. One arithmetic slip is noise; a pattern of generous marks is a process problem.
• Thresholds quietly restated between issues. Compare the same name's bar across the January, April and August updates.