Actionable insights — ETF Portfolio Update: February 2026

1. Separate the factor from the vehicle — a right idea can have a contaminated index

1. Establish the factor with long-run evidence: how much it has paid, over how long, and in what units.
2. Identify the default vehicle everyone uses to express it, and ask what that index actually admits.
3. Name the contamination specifically — here, the proportion of loss-making companies in a broad small-cap index.
4. Look for an index that filters the contaminant out before you accept a higher fee for it.

• Screening away the very stocks that produced the historical premium — much of the small-cap return has historically come from names that looked fragile at the time.

2. Stack screens in order of what they remove, and name the result

1. Start from the universe and apply a binary survival filter first — profitable for several consecutive years removes the tail without a judgement call.
2. Then apply the ranking filters: cash retained per dollar of revenue, and cash generated per dollar invested.
3. Cut to a fixed number of positions so the process is mechanical and repeatable.
4. Give the combination a name you can hold yourself to — "Small-High Quality" — so the strategy can be judged rather than drifted from.

• Every screen being backward-looking — seven years of past profits selects for businesses at the top of their cycle as readily as for durable ones.

3. Anchor a relative-valuation claim to its own long history, including when it ran the other way

1. Express the valuation as a ratio between two groups, not as an absolute multiple.
2. Chart it over a long window and state where today sits within that range.
3. Say explicitly when the ratio was inverted — a group that used to trade at a premium and now trades at a discount is a stronger claim than one that has always been cheap.
4. Note the direction of travel: a gap that has kept widening is not yet mean-reverting, and the trade may be early.

• Composition change behind the ratio — large-cap index earnings are far more technology-weighted than in 2004, so the two groups are not the same objects being compared over time.

4. Rank the five uses of free cash flow, and screen for the one you want

1. List what a company can do with surplus cash: repay debt, reinvest, acquire, pay dividends, buy back stock.
2. Decide which you are paying for and screen for it — reinvestment at high returns compounds inside the business and is not taxed on the way through.
3. Use return on invested capital as the test of whether reinvestment is creating value rather than merely consuming cash.
4. Be consistent about it, or say when you are not: elsewhere the same archive counts dividends and buybacks as return.

• Reinvestment with nowhere good to go — high ROIC on today's capital says nothing about the return on the next dollar.

5. Read a screened fund's top holdings before accepting its label

1. Pull the top ten and the sector split before buying any rules-based fund.
2. Check concentration: what share of the fund the top ten represent, and whether any single name is oversized.
3. Ask whether the businesses that came out of the screen match the story the screen tells — a "quality cash-generator" filter can surface cyclicals at peak cash flow.
4. Weigh the fee against how much of the effect you could get from a cheaper vehicle.

• Consumer lenders and commodity producers dominating a "quality free cash flow" screen — their cash flow is real and their cycle is not in the rear-view mirror.

6. Run paired portfolios when your audience cannot buy the same funds

1. Accept the access constraint as a structural fact and build two books that express the same factors through locally-available vehicles.
2. Make every transaction a matched pair, same size, same week, so the two books stay comparable.
3. Report both CAGRs side by side and resist explaining a divergence you cannot attribute.
4. Track which pairs diverge most — that is where the vehicles are not actually equivalent.

• Currency doing the work — the non-American book buys in euros and is marked in dollars, so part of the 4.4pp CAGR gap is FX, not selection.

7. Check that the transaction matches the argument

1. After writing the case, state what you actually bought and reconcile it against the case.
2. If you are not buying the vehicle you just argued for, give the reason and check that it covers every book you run.
3. Keep the ticket size and cadence fixed so the argument cannot inflate the position.

• The spotlight becoming marketing rather than allocation — an ETF of the Month that is never bought is a recommendation with no skin in it.