When to Sell a Stock

Why is selling a stock harder than buying one?

Buying is a positive act with an unbounded upside — the most you can lose is the position you sized, the most you can make is many times that. Selling is a closing act that crystallises both an outcome (the gain or loss) and a future regret (the gain you forfeit if the stock keeps running, or the loss you accept by exiting). The asymmetry makes the brain reach for any excuse to defer the decision, which is why most retail portfolios are run by a buying discipline and no selling discipline at all. The cost is enormous and largely invisible.

The asymmetry is also reinforced by tax. In a taxable account, the sell is a realisation event; the unrealised gain is "free" leverage on the position. The compounding logic alone is enough to bias the rational investor toward holding longer than the price would otherwise suggest. The combination of psychology and tax is why Buffett's preferred holding period — "forever" — is not a slogan but a tax-aware compounding argument.

What is Buffett's philosophy on when to sell?

Buffett's answer has evolved over six decades but the modern version is the cleanest: he sells almost never, because the businesses he owns are quality compounders whose intrinsic value rises faster than the price. He will sell if the business has been fundamentally impaired (the moat is gone), if a much better opportunity requires the capital, or if the original thesis turns out to have been wrong. Crucially, he does not sell because the price has run ahead of fair value in the short term — the tax friction and reinvestment risk of selling a quality compounder are higher than the recovery of a slightly stretched multiple.

The early Buffett of the 1960s was a different investor — a Graham-trained deep-value buyer who sold at intrinsic value because the businesses he owned could not be held forever. The transition to modern Buffett — buyer of See's, Coca-Cola, Apple — required a belief that the businesses being bought were structurally different. The selling discipline followed the buying discipline; both evolved together.

What is Peter Lynch's selling rule?

Lynch's rule was simpler and more retail-friendly: let winners run until the story changes. He famously categorised holdings into six buckets (slow growers, stalwarts, fast growers, cyclicals, turnarounds, asset plays), and each bucket had a different sell trigger. The unifying principle was that selling on price was almost always a mistake; selling because the business no longer fit the bucket was almost always right. Lynch's classic regret was selling Walmart at 8× earnings because it felt expensive — by his own admission, that one decision cost him more than most of his bad buys combined.

The Walmart story is worth quoting in full because it captures the single most expensive mistake retail investors make. Selling a compounder because the multiple looks high in the moment, even though the business is still doing what the buy thesis said it would, foregoes years of compounding that the price ahead-of-itself rationalisation cannot recover. The price will catch its breath; the business will keep working.

What is Klarman's defensive selling approach?

Klarman runs a defensive book at Baupost — capital preservation is the explicit first objective, not return. The Klarman rule is to sell at or near intrinsic value, because the margin of safety has been consumed and the asymmetric upside has flattened. The discipline produces lower returns in long bull markets and significantly higher returns through cycles. The honest comparison is not Klarman vs Buffett — the two are running different mandates. Klarman is solving for survivability; Buffett is solving for compounding inside survivable businesses.

The two philosophies are not in conflict; they are answers to different mandates. The retail value investor running a personal book has the option of either, but should pick one and stick to it. The investor who tries to be Klarman in the bull market and Buffett in the bear market ends up being neither, because the rule has become a rationalisation rather than a discipline. The choice of philosophy is itself the discipline.

When should I actually sell a stock?

Four conditions. First, the thesis has broken — the specific claim you wrote at entry is now false. Second, a kill criterion has tripped — a pre-written condition you set at entry has been met. Third, a clearly better opportunity requires the capital, after accounting for tax friction and the comparison must be specific (this name, not 'the market generally'). Fourth, the position size has drifted past your underwriting size and trimming is required for risk control. Any other reason is usually a discomfort talking, not the work.

Notice what is not on the list: price moves. The stock falling 30% is not a sell signal if the thesis is intact; the stock rising 50% is not a sell signal if the intrinsic value is still ahead of the price. Selling on price is the most common retail mistake, and it is the mistake the four-condition test is designed to prevent. If none of the four conditions has been met, the answer to "should I sell?" is almost always no — the question is your discomfort, not your work.

How do I sell without regret if the stock keeps rising?

You don't, fully. The regret of selling a winner that keeps running is the price you pay for the discipline of selling on process. The honest accounting is to track every sell decision against what the stock did over the subsequent two years. You will find that roughly half of your sells were vindicated by lower prices, a third look mistaken in hindsight, and the remainder are noise. The third that look mistaken are the cost of the discipline; the discipline itself produces returns the regret-driven alternative cannot.

The tracking exercise is genuinely useful. Most investors believe their sells are predominantly wrong because the wrong ones are memorable and the right ones are quiet. The data, when actually kept, is more balanced. Half of the sells you live with as regretful turn out to have been correct. The discipline is what produces the half that are correct; the regretful half is the price of the discipline. Across a long career, the trade is favourable.

Should I trim rather than fully sell?

Often yes. Trimming reduces position-size risk without forcing you to take a binary view on the thesis, which is rarely as black-and-white as a full sell implies. The honest rule is: trim when the position size has run past your underwriting size and you are no longer comfortable defending the full position; fully sell when the thesis has broken or a kill criterion has tripped. The two decisions are different. Conflating them produces over-selling on price moves and under-selling on broken theses.

Most retail portfolios benefit from a trim-first discipline because the alternative — wait for the binary decision — is usually deferred too long. The position runs past your underwriting size, the trim that should have happened at +50% does not happen, the position is now 12% of the book instead of the 6% you would have sized at entry, and the eventual fall hurts twice as much. The annual re-underwriting cycle (covered in the psychology post) catches this if you let it.

What this means for your watchlist

Pick a sell philosophy and write it down. Either you are a Buffett compounder-holder who sells only on broken theses, or you are a Klarman defensive seller who sells at intrinsic value, or you are running a hybrid where the rule is explicit about which bucket each holding sits in. Once the philosophy is written, the next sell decision becomes a question with a real answer instead of a feeling in search of a justification. That is the entire game.