Actionable insights — Portfolio Update: Buying more?

1. Trace a sector derating back to a specific product release, not a mood

1. When a whole sector reprices, look for a dateable event — a product launch, a regulation, a single earnings call — rather than accepting "sentiment" as the cause.
2. Describe precisely what the new thing does, and to whose revenue line. Here: a tool automating legal, sales and marketing tasks, which are seat-licence categories.
3. State the market's inference in one sentence so it can be tested: AI agents replace software licences, therefore seat counts fall.
4. Quantify the damage across named comparables over a fixed window, so the size of the repricing is on the record.
5. Then decide, business by business, whether the inference actually reaches that company's revenue.

• Naming the catalyst but never testing it per name — this issue lists four casualties and takes a view on none of them.

2. Publish an expected return with its assumptions separated

1. Write the projected return as an equation with named inputs: earnings growth, plus (or minus) the change in the multiple.
2. Show both numbers explicitly — here, 13.6% Owner's Earnings growth and a multiple moving from 17.1x to 22.0x.
3. Compute what the return would be on growth alone, so the reader can see how much rests on re-rating. Roughly a third of the 19.3% here comes from the multiple.
4. Keep a second, more conservative figure alongside it (the 15.0% "expected return for Our Portfolio") so the headline number is not the only one on the page.

• A multiple-expansion assumption presented as arithmetic — 17.1x to 22.0x is a forecast about other investors, not about the businesses.

3. Publish the profit and loss for every position, not just the winners

1. Chart unrealised gain and loss per holding in one image, sorted, with no aggregation to hide behind.
2. Name the winners doing the work, and let the chart show how few they are.
3. Leave the largest loss visible and unexplained if you have nothing new to say — the disclosure is the discipline.
4. Cross-check the picture against the ratings: a name deep in the red and not in the top rating tier is the one to interrogate.

• Charts without axis labels or a table behind them — approximate bars make the disclosure look fuller than it is.

4. Separate volatility from risk explicitly, and price the trade-off

1. State the alternative honestly: the risk-free return is the return you get for accepting no volatility.
2. Attach a number to what you are asking for — 10% a year from equities — so the volatility is being bought for something specific.
3. Make the position falsifiable by keeping it separate from the thesis: volatility being tolerable does not make any particular holding right.
4. Test yourself against your own disclosure — this page argues volatility is a fee while showing a −$44,000 position.

• The reframe used to avoid the question — permanent capital loss and a drawdown look identical for a long time.

5. Prefer growth that is already contracted over growth that is forecast

1. When a cheap stock needs future growth, ask what part of it is already signed rather than projected.
2. List the agreements with dates and expected revenue, so the growth has a schedule you can check against.
3. Separate the headwinds into structural and cyclical, and say which you expect to resolve without management action.
4. Only then look at the multiple; a low multiple on contracted growth is a different proposition from a low multiple on hope.

• Licence renewal risk — a business built on other people's brands has contracted growth and contracted expiry.

6. Reconcile this month's ratings against last month's, and explain the changes

1. Keep the Strong Buy list short and publish it in full each month.
2. Diff it against the previous list and state what moved and why — additions and removals both.
3. Reconcile the count of holdings against the disclosed weights; a mismatch means the sheet and the book have drifted apart.
4. Name the holdings that are not rated Buy — "7 strong buys and 8 buys" out of 18 leaves three unaccounted for.

• Silent upgrades — a name moving into the top tier without a stated reason is a rating you cannot audit.

7. Read position weight as a to-do list, not just a disclosure

1. Sort holdings by weight and compare that order against your conviction ranking.
2. Flag every name where the two disagree — a top-tier rating on a bottom-quartile weight is an unfinished decision.
3. Act on the gap deliberately, and say so when you do.

• Adding to a low weight purely because it is low — the size gap is a prompt to re-underwrite, not a reason on its own.