Actionable insights — ETF Portfolio Update: June 2026

1. Stack premia that are documented separately, and require both

1. Identify two return premia with independent long-run evidence and different causes — here a competitive-advantage premium and a size premium.
2. Check the causes really are different; two factors driven by the same thing are one bet.
3. Find (or build) an instrument that requires both conditions rather than blending them.
4. Quantify each premium and note the source, so the claim can be re-tested later.
5. Accept that requiring two conditions shrinks the universe — and check the universe left is still investable.

• Both figures come from the index providers whose products they justify — treat them as marketing-adjacent evidence.
• Premia measured from 2008 include a specific regime; the size premium in particular has long stretches of not working.

2. Classify the moat by source before you accept that it exists

1. For any company you call durable, name which of the five sources applies: switching costs, intangible assets, network effects, cost advantage, or efficient scale.
2. Write the specific mechanism in one sentence, naming the customer and what it would cost them to leave.
3. If no source fits, the advantage is probably execution — which is real but not durable.
4. Re-test the mechanism when the industry changes; a switching-cost moat can be dissolved by a new integration standard.

• Moats asserted from historical margins alone — high returns are the evidence of a moat, not the moat.
• Ratings agencies' "narrow" category being treated as the same thing as "wide" once it is inside a fund.

3. Test a growth forecast against the absolute money it requires

1. Convert the growth rate into an absolute amount of new revenue or profit.
2. Compare that amount to something concrete the company already owns — a division, a product line, a competitor.
3. Ask what would have to be invented or captured to produce it.
4. Apply the same test in reverse to a small company: the amount needed is often trivially achievable, which is the size premium in one calculation.

• The same test applied to your own holdings: BRK.B is used here as the example of growth slowing with size, one week after being upgraded to Buy.
• Large companies that escape the arithmetic through margin expansion rather than revenue — the test is about revenue, and it can mislead.

4. Time a factor with a relative-valuation ratio, not a forecast

1. Build one ratio: the valuation of the factor you want divided by the valuation of the thing it is measured against.
2. Plot it over the longest history you can get and mark the mean.
3. Act when the ratio is at an extreme, and size the action by how extreme it is — not by a view on when it reverts.
4. Add a structural reason the mispricing persists, so you know what would end it.
5. Keep buying on the schedule regardless; the ratio decides emphasis, not participation.

• Composition drift in the ratio — today's small-cap index carries more unprofitable companies than the 2004 version, which can justify a permanently lower multiple.
• A cheap ratio staying cheap for years; this one has been below average since roughly 2021.

5. Keep a transaction log that shows every purchase, including the losing one

1. Record date, instrument, quantity, price and amount for every purchase — never just the current weights.
2. Mark each line to market so the entry price and the result sit side by side.
3. Report the portfolio CAGR alongside, so the log and the headline number can be reconciled.
4. Leave the losers in the table. A log with no red line is a marketing document.

• A young log flattered by a strong start — the earliest 2023 entries carry most of the gain.
• Averaging into the biggest winner (VB, its fourth purchase) versus into the laggard: the log makes that choice visible, which is the point of keeping it.

6. Look through a fund's label to its actual holdings

1. Read the top holdings and the sector split before accepting the fund's description of itself.
2. Ask which screen produced them — a value filter applied after a quality filter will systematically surface cyclicals.
3. Check the concentration: what share of the fund is the top ten?
4. Decide whether the exposure you are getting is the exposure you wanted, and say so if it is not.

• Momentum screens inside "quality" products, which quietly make the fund trend-following at the margin.
• The same look-through on your own book: the archive's other portfolio ETFs are named only in chart labels, without tickers, which makes them harder to check.