Actionable insights — ETF Portfolio Update: Just Buy The Haystack

1. Bound an extrapolated trend by its own arithmetic

1. Take the trend everyone is extrapolating and ask what it implies if continued indefinitely.
2. Where the implication is impossible — a share of a total exceeding 100% — you have a hard ceiling, which is stronger than any forecast.
3. Then check whether the trend has ever reversed historically, using rolling multi-year windows rather than a single long chart.
4. Look for current evidence that the reversal has started, and separate that evidence from the ceiling argument.

• The ceiling being far away. "Cannot continue forever" says nothing about when, and this argument has been available and wrong for fifteen years.

2. Express every factor twice — once per accessible jurisdiction

1. Write down the factors you want exposure to (here: quality, size, multifactor, emerging markets) before choosing any funds.
2. For each factor, find both a US-listed fund and a UCITS equivalent, because "if you live in the US, you can't buy non-US ETFs. And if you live outside the US, you can't buy US ETFs."
3. Keep the two books at similar weights so the pair is comparable.
4. Match on the factor and the index methodology, not on the sponsor's name.

• Imperfect twins. IQLT is developed-international while IEQU is Europe-only; the top ten overlap, the tail does not.

3. Use paired funds as a natural experiment on geography

1. Chart each factor's US and non-US expression side by side over the same period.
2. Read the dispersion: a wide gap means the return came from geography, not from the factor.
3. Use that to decide where to add — the same idea at a much lower price is available on one side of the pair.
4. Set the honest expectation that the gap is not predictive: "over time, I expect both portfolios to generate similar returns."

• Currency. A euro-denominated twin and a dollar-denominated one are not measuring the same thing, and the issue does not adjust for it.

4. Publish the losing transactions, not just the aggregate return

1. Record every purchase with its date, price, quantity and current mark — not just the portfolio total.
2. Show the negative lines with the same prominence as the positive ones.
3. Revisit the reasoning given at the time of a losing purchase, so the process is reviewed rather than just the outcome.
4. Report the portfolio CAGR alongside, so the individual lines can be seen in proportion.

• A short window flattering or damning a decision. Six weeks says nothing about a small-cap valuation thesis either way.

5. Reduce a quality screen to three measurable inputs

1. Define quality by three things you can measure from filings: return on equity (profitability), earnings variability (predictability) and debt-to-equity (safety).
2. Apply the screen mechanically to a broad universe rather than case by case.
3. Then look at what it actually bought — the resulting top ten is the real definition of the screen.
4. Judge the screen by whether that portfolio is one you would want, not by the elegance of the rules.

• Trailing metrics buying cyclicals at peak earnings — the same caveat the archive raises against SPHQ's screen elsewhere.

6. Let the schedule, not the scoreboard, choose the position to top up

1. Decide in advance whether monthly additions go to the target-weight gap, the worst performer, or the strongest conviction — and write the rule down.
2. If the rule is weight-based, it will systematically add to laggards, which is the intended behaviour, not a mistake.
3. State whether any given purchase is the rule executing or a judgment overriding it.
4. Review annually whether the laggards recovered — that is the test of the rule, not of any single purchase.

• A rule quietly becoming a judgment. Adding to the laggard is defensible; adding to the laggard and then rationalising it afterwards is not the same thing.