Sell Your Winners, Marry Your Losers: The Disposition Effect

Ask any broker who has watched retail clients for twenty years what the single most reliable pattern in human trading is, and you will get the same answer. People sell their winners and hold their losers.

They will tell you this as folk wisdom, a sad joke about human nature. In 1985, Hersh Shefrin and Meir Statman took the joke seriously, gave it a name, built a theory for it, and showed it was not a joke at all. They called it the disposition effect, and it is arguably the most robust behavioral finance finding in existence: it shows up in every country, in every asset class, and in every decade anyone has bothered to check.

The problem: the tax code says do the exact opposite

Here is what makes this so damning. The disposition effect is not just psychologically interesting, it is financially insane, and provably so, on grounds the investor himself would agree with.

Consider the US tax system (and most others). If you sell a stock at a loss, you can use that loss to offset gains elsewhere and reduce your tax bill. If you sell at a gain, you create a tax bill.

So the tax-optimal behavior is obvious and unambiguous: realize your losses, defer your gains. Sell losers, hold winners. It is the exact opposite of what everyone actually does.

This is what makes the disposition effect such a beautiful piece of evidence. You cannot explain it away as some clever rational strategy, because the most obvious rational consideration in the room, the tax code, is screaming at people to do the reverse and they still do not listen. Whatever is driving this, it is strong enough to overcome free money from the government.

Shefrin and Statman also examined whether it might be an information story: maybe people sell winners because they rationally believe winners are about to mean-revert. The problem is that if they truly believed the losers were going to bounce back and the winners were going to fall, then the correct move is not just to hold the losers, it is to buy more of them. People do not do that either. They just sit there, frozen, refusing to sell.

The key idea via analogy: closing the file

Shefrin and Statman explain the effect by assembling several psychological ingredients, and the central one is mental accounting joined to prospect theory.

Picture each stock you own as a folder on your desk. When you buy a stock, you open a folder with the purchase price written on the front. That price is now the reference point for that folder, and everything that happens is scored as a gain or a loss relative to it.

Crucially, the folder does not close until you sell. Selling is the act of closing the file and writing the final result on it in permanent ink.

Now apply prospect theory, which tells you two things:

• In the domain of gains, people are risk-averse. The stock is up 20 percent. Closing the folder means booking a certain win. Holding means risking that win. Prospect theory says grab the sure thing. So you sell the winner, cheerfully, and enjoy the small hit of pleasure that comes from a folder marked "success".
• In the domain of losses, people are risk-seeking. The stock is down 20 percent. Closing the folder means writing "loss" on it forever, which is a certain, permanent, painful admission. Holding means there is still a chance it comes back to break even and the folder can be closed as a draw. Prospect theory says gamble to avoid the certain loss. So you hold the loser, indefinitely.

The stock does not know or care what you paid for it. Its future returns have nothing whatsoever to do with your cost basis. But your folder cares enormously, because your cost basis is where you drew the line between winning and losing.

Shefrin and Statman layered on three more ingredients that make the effect stickier:

• Regret aversion. Selling at a loss is not just financially painful, it is an admission that you made a mistake. As long as you hold, the story is unfinished and you were merely early. The moment you sell, you were wrong. People will pay a lot of money to avoid that sentence.
• Self-control. You know you should cut the loser. You cannot make yourself do it. This is the same structure as dieting or quitting smoking: a planner who knows the right answer and a doer who will not comply. The remedy is the same too, which is a rule that removes the decision from the moment, in other words, a stop loss.
• Mental accounting stops you netting things out. Because each stock lives in its own folder, you do not naturally think "my portfolio is up, so this individual loss is irrelevant". Each folder demands its own happy ending.

Why it mattered

• It named a bias that costs real people real money, in a way they can see. Most behavioral findings are abstract. This one, you can check on your own brokerage statement in about ninety seconds. Sort your realized trades by profit, look at your average holding period for winners versus losers, and prepare to feel bad.
• It set up the empirical work that proved it beyond argument. Terrance Odean's 1998 paper took thousands of real brokerage accounts and confirmed the effect cleanly and unmistakably in the data, and even showed the winners people sold went on to outperform the losers they kept, meaning the behavior is not merely tax-inefficient, it is directly return-destroying. Shefrin and Statman supplied the theory. Odean supplied the smoking gun.
• It explains why stop losses exist, and why nobody uses them. A stop loss is a machine for defeating the disposition effect: it forces the folder closed before your risk-seeking loss-domain brain gets a vote. This is also exactly why people move their stops, which is the disposition effect defeating the machine.
• It has consequences for prices, not just for individuals. If large numbers of investors refuse to sell losers, then selling pressure on falling stocks is artificially suppressed, so bad news gets into prices too slowly. That is a plausible micro-foundation for momentum, and later researchers have used unrealized gains and losses in the investor base as a genuine predictor of returns.

The honest limitations

• The original evidence was thin. The 1985 paper is primarily a theory paper. Its direct evidence was suggestive rather than overwhelming, using aggregate patterns and some mutual fund data. The overwhelming evidence came later, from account-level datasets that did not exist yet when they wrote it. They were right, but they were partly right on faith.
• There are non-behavioral explanations, and they are not all silly. An investor might rationally hold a loser if he believes in mean reversion, or if he faces transaction costs, or if he is rebalancing toward a target weight (which mechanically means buying what fell). Disentangling a stubborn ego from a disciplined rebalancing rule is harder than it looks.
• It reverses in December. The disposition effect notably weakens or flips at the end of the tax year, when the tax incentive finally becomes salient enough to overpower the psychology. That is fascinating, and slightly humiliating for humanity: we are capable of doing the rational thing, but apparently only when there is a deadline.
• Sophisticated investors do it less. The effect is strongest in retail and weakens among professionals and experienced traders, though it does not vanish. This is comforting if you are a professional and inconvenient if you want to build a trading strategy around exploiting it.

The one-line takeaway

Shefrin and Statman explained why investors sell winners too soon and cling to losers too long: each position is a mental folder scored against its purchase price, and closing a losing folder means permanently admitting a mistake, so people gamble on a recovery that the stock, which has never heard of your cost basis, has no obligation to deliver.