Actionable insights — ETF Portfolio Update July 2026

1. Decompose every past return into cash growth and multiple change before drawing a lesson from it

1. For any share price move, obtain the change in free cash flow (or earnings) and the change in the multiple over the same window.
2. Multiply them: the return is (1 + cash growth) × (1 + multiple change) − 1. They compound, they do not add.
3. Attribute the result. A return driven by cash growth is repeatable if the business keeps growing; one driven by multiple expansion is borrowed from future returns.
4. Run the same decomposition on your own winners. If the multiple did the work, your thesis was not confirmed — you were paid for a change in sentiment.
5. Use it forward as well as backward: state which engine you expect to fire before buying, so you can check afterwards whether you were right for your reason.

• Choosing endpoints that flatter the decomposition; use full cycles where possible.
• Multiple expansion being mistaken for validation — it is the component most likely to reverse.

2. Convert the share price into the whole-company price before you buy

1. Multiply the share price by the share count to get what the market is asking for the entire business.
2. Ask the question literally: would you buy this whole company for that sum, if you had it?
3. Sanity-check the number against something concrete — the annual cash it produces, or what comparable private businesses change hands for.
4. Repeat it for a position you already own; familiarity hides the size of the claim you are making.

• Market capitalisation ignoring debt — for a leveraged business, enterprise value is the number you would actually pay.
• The test becoming rhetorical; it only works if you are willing to answer no.

3. Trade a factor spread at an extreme rather than forecasting the market

1. Choose a pair with a long, stable relationship — high versus low volatility, value versus growth, small versus large.
2. Chart the relative performance over decades and identify how far the current reading sits from its range.
3. Express the finding as a valuation statement about the cheap leg, not as a prediction about the market's direction.
4. Act in small, scheduled increments; a spread at an extreme can widen for years before it closes.
5. Define in advance what closes the position — a return of the spread to its median, not a target price.

• The 2006 comparison being an ominous precedent rather than a comforting one; what followed was not a gentle mean reversion.
• Composition drift — the constituents of a "low volatility" basket in 2026 are not those of 2006.

4. Judge a rules-based fund by what its rule actually bought

1. Read the methodology, in order: which universe, which filter first, which filter second. Order changes the output.
2. Then read the actual top holdings and sector weights, which are the rule's confession.
3. Ask whether the names you find match the description you were sold. A "quality" screen holding deep cyclicals is telling you what its metric really selects for.
4. Check concentration — the share of the fund in its top ten — so you know whether you are buying a portfolio or a handful of bets.
5. Treat sponsor backtests as marketing until independently verified; note the source explicitly in your own notes.

• Cyclical cash flow at a peak scoring as "cheap" — the classic failure mode of a free-cash-flow-yield screen.
• Rebalancing frequency: a screen that reconstitutes annually can hold what it would no longer select.

5. Direct new money to the weakest position in the plan, not the strongest

1. Set target weights for the portfolio in advance and record them.
2. Each month, compare actual weights with targets and identify the largest shortfall.
3. Direct the new contribution there, provided the reason for the underperformance is price rather than a broken thesis.
4. Write down which of the two it is, every time — this is the only step that separates rebalancing from averaging down into a mistake.
5. Keep the increments small and regular so the decision is never large enough to require confidence.

• Mechanical addition to a position whose underlying case has genuinely deteriorated.
• A "target weight" that quietly moves to justify the purchase.

6. Build one strategy with two implementations when access, not conviction, differs

1. Separate the strategy (the factor exposures you want) from the instruments (the funds you can legally and practically buy).
2. Where regulation, domicile or tax blocks an instrument, find the closest available equivalent rather than changing the strategy.
3. Keep the factor weights aligned across implementations so the two remain comparable.
4. Track them side by side, and expect divergence from instrument differences rather than reading it as evidence about the strategy.

• Currency exposure differing between implementations, which can dominate short-run comparisons.
• Fee and tax-treatment differences quietly making one implementation structurally worse.