When to sell a stock: thesis broken vs a temporary price drop

Knowing when to sell a stock is not a chart problem. It is a thesis problem. The quote is a vote; the thesis is the argument you actually underwrote.

The quote is not the thesis

The search “when to sell a stock” is usually typed after a red print. That is backwards. A moving quote tells you something about other people's positioning, liquidity, and mood. Sometimes it also tells you something about the business. Those two signals get collapsed because the quote is the only number that updates every second, and the business updates on a slower clock — filings, customers, credit, competition.

If you entered because a story felt exciting, you have nothing to check when the price falls. You will invent a new story, or you will sell to stop the discomfort. Either way you are trading your nervous system. If you entered with a written thesis — what has to be true, what would prove you wrong, over what horizon — the drop becomes a question you can actually answer: did the facts change, or only the vote?

We have written about that split in the early life of a live book: separating thesis broken from price noisy. The same test applies after year one. Time does not turn a quote into a thesis.

A four-state map, not a percentage rule

Generic “3 reasons to sell” lists usually smuggle in a price rule: down 20%, take the loss; up 50%, take the gain. Those are not investment rules. They are ways to make the P&L feel tidy. A long-term book needs a map that can hold four different situations without pretending they are the same trade.

Notice what is missing: a stop-loss percentage, a moving average, an alert. Those tools answer “when did the chart move.” They do not answer “is the business we underwrote still the business we own.”

Price down, thesis intact

This is the case that feels like a test and usually is. A quarter misses because of weather, a one-off legal reserve, or a cycle the original note already named. Credit is still the credit you underwrote. The competitive position has not flipped. You would still initiate today at this price, size held constant.

The disciplined move is not heroic buying. It is not selling for the sake of feeling active. Adding is optional and constrained: only if the position is still inside your size rules, and only if you are adding to the same thesis, not to a new one you wrote after the drop. Averaging down is correct when the price improved and the facts did not deteriorate. It is expensive when you are negotiating with a company that is no longer the one you bought. For how that sits inside a concentrated book, see how many stocks you should hold.

Price down, thesis weakening

Weakening is the gray band people skip because it does not feel decisive. Churn is edging up. Guidance quality is getting vaguer. A customer concentration you tolerated is becoming the story. None of that is yet a full invalidation. All of it is a reason to stop treating the position as automatically “a hold.”

The useful move is a trim plus an evidence deadline. Write down the fact that would restore the original underwrite, and the date by which it should be visible — an earnings print, a filing, a contract, a credit metric. If the date arrives and the fact does not, you have converted fog into a broken thesis. If you refuse to name the date, you are hoping the quote will make the decision for you.

Thesis broken — exit regardless of P&L

Broken means the reason you own the shares is gone. The moat you underwrote was a regulation, and the regulation changed. The balance sheet you needed is now the risk. The product cycle you were paying for was delayed into a different company. At that point the cost basis is a historical curiosity. Selling a loser does not make you “wrong twice.” Holding a thesis that no longer exists is how a small mistake becomes a character trait.

This is also true when you are up. A gain does not repair a dead argument. “I'll wait to get back to even” and “I'll wait until it rounds out a nice percentage” are the same error with different vanity. The exit is about the business, not the souvenir.

Should you find yourself in a chronically leaking boat, energy devoted to changing vessels is likely to be more productive than energy devoted to patching leaks.

Price up or flat, thesis rotting

This is the case almost nobody searches for, and the one that does the most damage in a long-term book. The tape is calm. The position may even be a winner. Meanwhile unit economics are slipping, the capital cycle is turning against you, or management is quietly changing the business you thought you owned. Because there is no red number, there is no meeting with yourself.

Price is a lagging auditor. If you only review names that hurt, you will keep the rotting ones until the quote finally agrees — at which point you will sell into the recognition, which is the expensive sequence. A biweekly or monthly review that starts with the original note, not the P&L is how you catch this before the market does the work for you. That is the same slow cadence we argued for in how to beat the S&P 500 without becoming a day trader: thirty minutes on theses, not a day on candles.

What an exit rule actually is

An exit rule is a sentence you can falsify. It is not a feeling, and it is not a chart setup. The useful form is: we would be wrong if X is no longer true, evidenced by Y, over horizon Z. X is the economic claim. Y is an observable — a margin, a credit ratio, a customer metric, a regulatory outcome. Z is long enough for a business to speak and short enough that you cannot hide.

That is different from a stop-loss. A stop-loss says: if enough other people sell, I will too. Sometimes that coincides with a broken business. Often it coincides with a noisy quarter, a sector flush, or a forced seller. Long-term underwriting is the decision to live with that noise on purpose, which only works if you wrote down what noise is allowed to look like.

Bad quarters are not a broken thesis. We underwrite over years. A business can miss a print, sit through a down cycle, and still be the company in the original note. The error is treating every ugly month as a verdict, or treating every calm month as permission to stop reading. Pre-defined exits belong in the entry, not in the panic. That is one of the five boring rules in how to outperform the S&P 500 with stock picks: you exit when the reason you bought is no longer true.

How Outpick treats selling

Most “when to sell” posts are trying to be a trading desk in essay form. We are not. Outpick is a research firm that publishes high-conviction ideas on a slow cadence. The sell discipline is part of the research, not a separate product.

• A full thesis, including how we would be wrong. Every initiation note should state the economic claim and the facts that would invalidate it. If that paragraph is missing, you do not have an exit rule. You have a hope.
• Exit notes when a position closes — including losers. A closed name is not a deleted name. We write why we left. That is the only way a track record can be audited as a process instead of a highlight reel. Closed work lives on the public track record; the point is the reasoning, not a price alert.
• Not a signal service. We do not send entry and exit prices as trade instructions. We do not publish chart setups, stop levels, or “sell now” pings. Members get the argument. They size and execute for their own accounts, on their own timetable.
• Years, not weeks. The book is built to hold businesses long enough for a thesis to play out. A bad quarter is an input. It is not, by itself, a broken thesis. Cadence exists so we are not forced to invent a new opinion every session — see why we publish one stock pick every two weeks.

That is also how you should evaluate anyone selling stock research. A written thesis per pick, losses shown with the same prominence as wins, and a defined cadence are the minimum. We laid that checklist out in best stock-picking newsletters for long-term investors. Without an invalidation sentence, you cannot tell whether a winner was skill or luck, or whether a thesis has broken when the facts change.