Actionable insights — Portfolio Update: July 2026

1. Read a commodity peak backwards: from the losses that caused it

1. Establish whether the product is differentiated. The test is a question, not a ratio: could this company lose money because customers demanded lower prices? If yes, it is a commodity and its profits are a function of supply and demand, not of quality.
2. Trace the supply history. Extreme current profits usually follow a period of losses severe enough to have stopped capacity being built — find that period and quantify it.
3. Find the previous margin peak and look at what happened to margins, and then to the share price, in the following one to two years.
4. Compare today's margin with that prior peak. If it is higher, the invitation to competitors is stronger, not weaker.
5. Ask what the current price implies: is the market capitalising peak profits as if they were normal? If so, the bet you are being offered is "no mean reversion," and you can decline it without shorting.

• Genuine structural change in the demand curve — the cycle argument is a base rate, and base rates occasionally break.
• The lag: capacity takes years to build, so an over-earning cycle can run far longer than the logic suggests.

2. Apply the same cycle logic to your own holdings — and let a cost advantage change the answer

1. When you make a cyclical argument against something, immediately ask which of your own positions is in an equivalent price cycle.
2. For each, identify whether there is a structural cost or capability advantage that survives the down-leg. Commodity cycle plus no advantage is a pass; commodity cycle plus a durable low-cost position is an entry.
3. Confirm the mechanism explicitly: weaker competitors lose money, withdraw capacity, and prices recover — with your holding still standing.
4. Then test whether the market is pricing the down-leg as permanent. That, not the cycle itself, is the opportunity.
5. Write both cases side by side so the asymmetry is visible and can be challenged.

• An advantage that is really a cycle artefact — low loss ratios during benign years are not the same as a lower cost of doing business.
• Soft-market growth: an insurer expanding while prices fall may be buying share with underwriting standards.

3. Run a relative multiple and a reverse DCF, and require both to agree

1. Compare the forward multiple with the company's own long-run average, not with a sector or index figure. Express the gap as a percentage.
2. Separately, run a reverse DCF: at today's price and a 10% required return, what annual free-cash-flow growth is implied?
3. Compare that implied rate with the growth actually delivered over five and ten years.
4. Act when both readings point the same way. When they disagree, the disagreement is the finding — one of the two inputs is wrong.
5. State both numbers in the write-up so a reader can reject either independently.

• A historical average multiple struck over a period when the business was structurally different (higher growth, different mix).
• Reverse DCF hurdles that look trivial because the required return assumption is low; hold it constant at 10% across names.

4. Underwrite the whole portfolio the way you underwrite one stock

1. Aggregate the book into three or four numbers: weighted forward multiple, expected revenue growth, and the discount or premium versus the index.
2. Build the expected return from disclosed inputs — assume owner's earnings grow at the revenue rate (deliberately conservative), then assume a re-rating to a stated target multiple.
3. Report the sum, not a target. The reader should be able to disagree with one input and redo it.
4. Add a growth-of-intrinsic-value figure so the reader can see the portfolio's engine independently of its price.
5. Revisit the same four numbers each period; the trend in them is more informative than any single reading.

• A weighted average dominated by one or two large positions; check the median as well.
• The re-rating assumption quietly carrying the whole return — at a flat multiple, 6.8% growth is 6.8%, not 10%.

5. For a company buying back stock, treat share-price volatility as an input, not a cost

1. Identify holdings that retire a meaningful share of their own equity each year.
2. For these, invert the usual reading: a lower price means the same buyback budget retires more shares, permanently increasing each remaining holder's claim.
3. Confirm management actually buys into weakness rather than steadily, and check the share count trend rather than the dollars spent.
4. Verify the buyback is not merely absorbing stock-based compensation.
5. Adjust your own behaviour accordingly: a drawdown in a cannibal stock is a reason to hold or add, not to reassess.

• Buybacks funded with debt into a cyclical downturn — the mechanism reverses violently.
• A price fall that reflects a deteriorating business; the inversion only holds while the thesis does.

6. Keep a divergence case study on hand for the years the price is wrong

1. Collect examples of a share price falling for a year or more while the underlying free cash flow rose — ideally in a company you do not own, so the example is not self-serving.
2. Identify the fear that caused the divergence, and whether it eventually proved true, partly true, or false.
3. Use the case to set your own expectations of duration: divergences of two to three years are normal, not anomalous.
4. Convert it into a monitoring rule — track the cash flow line, not the price line, and define in advance what deterioration would count as a broken thesis.

• Survivorship in the examples — for every AbbVie there is a business where the market's fear was correct and the cash flow followed the price down.
• Using the case to defer a genuine thesis review; a divergence story is not a substitute for re-underwriting.

7. Pick a single public proxy for the factor you are exposed to, and watch it

1. Choose one liquid, well-understood security that behaves like the style you run, so you can observe the factor without buying a factor index.
2. Track it over two windows — a long one for where you have been, a short one for any turn.
3. Read a divergence between the windows as tentative evidence, and say so; do not size a position on it.
4. Pair the proxy with a breadth reading (the riskiest versus the safest cohorts relative to the index) so a single stock is not carrying the inference.

• One-month readings that are noise; require confirmation before treating the turn as real.
• Company-specific news in the proxy contaminating the factor signal.

8. Distinguish the two kinds of speculation running at the same time

1. Separate the crowd's enthusiasms into loss-makers priced on possibility and over-earners priced on a peak. They fail differently and on different timetables.
2. For the loss-makers, the question is funding and dilution; for the over-earners, it is capacity and mean reversion.
3. Note which of your own holdings would be affected by each unwinding, and how the two would interact.
4. Resist a single "bubble" label — the two groups are not the same trade and will not reverse together.

• Correlation in the unwind even though the causes differ — both groups depend on the same risk appetite.
• Your quality book being sold to fund either trade, which is what creates the entry points in the first place.