When to sell — the most underwritten topic in investing

Count the books on any investing shelf: fifty on what to buy, perhaps half a chapter on when to sell. Yet selling is where returns actually get decided. A well-chosen purchase held through a broken thesis, or sold three years too early, produces mediocre results either way. This is my framework, developed across seven years of public investing — including from my own mistakes. And it isn't theoretical: every position I've closed is on the record, with its return and how long I held it.

Four reasons I consider legitimate

The thesis broke. Not "the stock fell" — the specific, factual reason I bought is gone. This only works if the reason was written down at purchase, together with its falsifier: the observable event that would prove me wrong. Future-me is an unreliable negotiator, and a falsifier written when calm cannot be argued with when a position is down 30 percent.

The price reached what the math supports. I buy cash flows at a yield; when the price rises far enough, that yield stops compensating me. A position bought at a 9 percent free-cash-flow yield that now trades at 3 percent is paying me less than a government bond, with more risk attached. Nothing went wrong — the market simply finished agreeing with me.

A clearly better use of the same capital exists. This one carries a deliberately high bar, because switching has costs: taxes, spreads, and the well-documented risk that the new idea is simply newer, not better. The replacement has to be clearly superior after those frictions, not marginally more exciting before them.

The position grew too large. A winner that compounds into 25 percent of the portfolio has made the portfolio's fate depend on one name. Trimming a great business feels wrong every single time, which is exactly why the rule exists in advance rather than being decided in the moment.

Four reasons that feel legitimate but aren't

The price fell. By itself, a falling price contains no information about the business — it tells me what other people paid, not what the company earned. If the thesis is intact, a lower price is the same asset at a better yield.

Boredom. Some of the best holdings do nothing noteworthy for years, which is roughly the point of owning them. A red quarter within an intact thesis: quarters are noise, and any thesis that cannot survive one bad quarter was not a thesis. And "locking in gains": selling converts paper compounding into a permanent tax haircut. Sell a position that has doubled, pay 25 percent capital gains tax, and the replacement idea must return roughly 14 percent just to get back to where the old one already stood.

Decide calm, execute mechanically

The uncomfortable truth about selling is that the moment of decision is the worst possible moment to be deciding. A position down 40 percent generates fear; a position up 300 percent generates both greed and vertigo. Neither state produces good analysis.

So the rules get written in advance, at purchase, when nothing is at stake emotionally. What I bought, why, at what yield, and precisely what observable fact would break the thesis. When that fact arrives, I sell — not after one more quarter, not after the price recovers to my entry. When it does not arrive, I hold, whatever the chart looks like.

Two ways to lose

Selling too early costs quiet fortunes. Nobody computes the compounding they surrendered by exiting a great business in year three of thirty, so the loss never appears in any statement — it is simply money that went to someone more patient.

Holding broken theses costs loud ones. These losses do appear in the statement, usually in red, usually after months of hoping. The falsifier discipline is the answer to both errors at once: it forces the sale when the facts break, and it blocks the sale when only the price did. Everything else in this article is commentary on that one habit.