Actionable insights — Buying more of this stock

1. Open a new position at half size, then complete it with an identical second order

1. Decide the full target weight before the first purchase, then buy half of it.
2. Write down the conditions for the second half: time held, further work done, and the thesis unchanged.
3. Place the second order on the same terms as the first (same share count, same limit), so the add is not a reaction to price.
4. Announce the add as the completion of a plan, and point back to where the plan was made.

• Whether the other two 1 September "up" names get the same treatment. SPGI sits at ~2.85%, and a plan carried out for only one of three names is closer to a single trade than a policy.
• An unchanged limit after the price has moved well away from it: that means the order may not fill, and the post should say so.

2. Put the expected-return sheet next to the weights, and explain every mismatch

1. Rank the holdings by modelled expected return, and separately by current weight.
2. Flag the large positions with low expected returns and the small positions with high ones.
3. For each flag, write one line: why the size is right despite the number (quality, certainty, liquidity, tax), or why it should change.
4. Direct new money to the high-return, small-weight positions first, unless a flag's written reason says otherwise.

• Low-return positions that stay large issue after issue. At that point the weights reflect past gains, not current conviction.
• Outlier inputs at the top of the sheet: KPG's 122% three-year NPATA growth is a single forecast that can dominate the average.

3. Check that every row of a comparative model uses the same metric

1. Read the footnotes on any model table before reading the ranking.
2. List every row that uses a substitute metric (NPATA, revenue, book value, distributable earnings).
3. For each substitute, ask whether it grows at the same rate as per-share earnings for that business. If not, take the row out of the ranking.
4. Look for identical values across unrelated rows, which usually means a plug value rather than a forecast.

• The substituted row being the one that drives a transaction. That is when a like-for-like check matters most.
• A change in the book-level average from one issue to the next with no stated cause (17.1% on 1 September, 13.9% here).

4. Treat heavy buybacks as a management valuation signal, but check the size and the price paid

1. Pull the share-count history: the multi-year CAGR plus the most recent month or quarter.
2. Read an acceleration in buybacks as management saying the shares are cheap. Confirm it by comparing the average price paid with book value or intrinsic value.
3. Check the buyback is funded by surplus capital, not by leverage or at the cost of the core business (for an insurer, reserve strength).

• The average repurchase price relative to book value per share. The post does not give it.
• Buybacks slowing: if management stops buying at similar prices, one of the three stated reasons weakens.

5. In a style drawdown, report the cash yield and restate conviction instead of changing style

1. Say plainly that results are below expectations.
2. Replace the price chart with a cash measure: the portfolio's FCF yield against its own history.
3. Only buy names you could explain yourself. Conviction you have borrowed from someone else will not last through a drawdown.

• A rising FCF yield caused by falling prices rather than rising cash flow. The chart alone cannot tell the two apart, so check absolute look-through FCF (reported as $106,119/yr on 1 September).