We have published second reports on eight companies. Six of the eight scores went down. Two stayed flat. None went up. And not one of the eight became a sell.
That looks like cowardice until you see which way the prices moved. Microsoft's score fell from 8.3 to 7.6 while the shares rose 35.6%. Nike's fell from 6.2 to 5.4 while the shares fell 18.4%. Adobe's fell from 7.4 to 6.8 while the price moved nine cents.
Which means the score and the price are not measuring the same thing, and the sell decision lives in the gap between them.
Eight revisits, graded in public
| Price move | Our score | What we concluded | |
|---|---|---|---|
| Microsoft | +35.6% | 8.3 → 7.6 | Hold; do not chase |
| Nvidia | +12.5% | 7.8 → 7.6 | Still too hard |
| Coca-Cola | +9.6% | 8.0 → 7.8 | Accumulate, slowly |
| Adobe | +0.04% | 7.4 → 6.8 | Accumulate |
| Alphabet | −3.8% | 8.0 → 8.0 | Accumulate |
| Realty Income | −3.9% | 7.0 → 7.0 | Own for income |
| McDonald's | −4.4% | 7.8 → 7.3 | Buy the landlord — still |
| Nike | −18.4% | 6.2 → 5.4 | The zone arrived; the evidence didn't |
Read the first and last rows together. The best price performance came with the second-largest score cut. The worst price performance came with the largest. There is no relationship between the two columns, and that is not noise — it is the structure of the problem.
A share price tells you what other people will pay. A thesis tells you what you own. Selling on the first while ignoring the second is how good businesses get sold cheap.
Two axes, four squares
Price is one question. The business is a different one.
| Thesis intact | Thesis weaker | |
|---|---|---|
| Cheaper than when you bought | Add. The easiest square and the rarest — McDonald's at a fresh 52-week low with the franchise-landlord model untouched. | The hardest square. Nike: 18% cheaper and two of three tests failed. The price is arguing for you, which is exactly why it needs the most scrutiny. |
| Dearer than when you bought | Hold. Do not chase. Microsoft: better business, much better price, worse investment at the margin. A reason not to add, not a reason to leave. | The only clean sell. Paying more for less. If both axes have moved against you, the argument for staying is habit. |
Three of the four squares are not sells. That asymmetry is deliberate, and it is the reason our record has six score cuts and no exits. Selling costs you the compounding, the tax, the spread, and the requirement to find something better — so the bar should sit higher than "it went down" or "it went up a lot".
Our own, published in full
In June we called Microsoft a wonderful business at a fair price — and then asked for a bargain. Two days later the shares touched $349.20, inside our own estimate of what the company was worth, and we were waiting for $310.
They were $497.93 when we wrote the follow-up. We were right about the business and wrong to be greedy about it.
The lesson is not "pay any price". It is that a demand for a bargain is a decision, with a cost, and the cost is paid silently. The same reflex that stops you buying a good business too dear will stop you buying it at all.
PayPal is the mirror image. Our report found it 82% below its 2021 peak — cheap, and genuinely not a trap, with Stripe having bid $53 billion for it and the board saying that was too low. And our conclusion was that the bargain got away three weeks earlier.
Cheap enough, right about the business, three weeks late. That is the ordinary experience of investing, and it is worth internalising because it stops you treating every missed entry as a failure of analysis.
How to make the decision actually decidable
Not "great company, good moat". Something that can be graded: margins improve without a one-off benefit; this region stops shrinking; the order book grows faster than revenue. If you cannot write it down, you have a feeling rather than a thesis, and a feeling cannot be broken — which means you will never know when to sell.
Set the date when you buy — the next annual report, or four quarters out. Grading triggered by the share price means you only ever re-examine after the market has already formed its view, which is precisely when your judgement is worst.
These are two different decisions and merging them causes most of the damage. Microsoft after a 36% rise is a stop-adding. Nike after two failed tests is a stop-adding with a question attached. A sell requires both axes to have moved, or a clearly better use for the money.
"Sell" is only half an instruction. If you cannot name what the proceeds are going into — a specific alternative, or deliberately held cash with a stated purpose — you are not making a portfolio decision, you are reacting to discomfort.
Before selling, write the sentence you would need to have written on the day you bought, for selling today to be consistent with it.
If that sentence is "I will sell if the shares fall 20%", then you bought a price. If it is "I will sell if the margin does not recover without a one-off benefit", then you bought a business, and the September report either contains the answer or tells you when it will.
One we held while it rose 36%, one we kept at a fresh low, and one 18% cheaper with two tests failed.
Including the follow-ups where we graded our own calls in public.
Frequently asked
When should you sell a stock?
When the reason you bought it is no longer true, or when you find something clearly better for the same money. Neither of those is a price. A falling price is a reason to re-read the thesis, and a rising price is a reason to check what you are now being asked to pay — but the sell decision itself is about the business, not the quote.
Should I sell a stock that has gone up a lot?
Not merely because it went up. Our Microsoft score fell from 8.3 to 7.6 while the shares rose 35.6%, and the conclusion was still to hold rather than sell — a more expensive wonderful business is a worse investment than it was, which is a reason not to add rather than a reason to leave. Selling a compounding business for being expensive is how most long-term returns get given away.
Should I sell a stock at a loss?
Only on the evidence, never on the loss itself. The hardest square in investing is a share that is both cheaper and worse — our Nike score fell from 6.2 to 5.4 while the shares fell 18.4%. Cheaper and worse is not automatically a sell either; it is the case that requires the most work, because the price is doing some of the arguing for you.
How do I know if my investment thesis is broken?
Write down the two or three things that would have to stay true, before you buy, and grade them out loud later. A thesis is not broken because the price fell — it is broken when a specific thing you relied on stops being true. We set three tests for Nike in July and published the grades in September: one passed, two failed.