Execution · Selling
The Disposition Effect
Investors sell winners too early and hold losers too long. The pattern is robust, expensive, and mostly invisible to the person doing it.
What it is
Hersh Shefrin and Meir Statman named it in 1985. Prospect theory predicts that people are risk-averse when they are ahead and risk-seeking when they are behind, relative to a reference point. For most investors the reference point is the purchase price. Selling above it feels like a win. Selling below it feels like admitting a mistake. So winners get realised and losers get nursed.
What the brokerage data showed
Terrance Odean examined roughly 10,000 household accounts at a US discount broker from 1987 to 1993. Outside December, the proportion of gains realised was about 14.8 per cent against 9.8 per cent for losses. A winning stock was substantially more likely to be sold on a given day than a losing stock. The prior winners that were sold went on to outperform the prior losers that were held, by about 3.4 per cent over the following year in his sample. The disposition was not merely untidy. It was costly.
The pattern reversed in December, when tax-loss selling dominated. That reversal is useful: it shows the rest of the year is psychology, not optimisation.
Why it persists
Mental accounting treats each position as its own story with its own break-even. Loss aversion makes confirming the loss aversive. There is also a thin rationalisation available in Australia: deferring a sale can defer CGT and preserve franking. The rationalisation does not rescue a position whose thesis has already failed. It just makes the hold feel prudent.
Institutions show the effect less strongly, but it appears where performance is judged frequently and career or redemption risk is salient. It is not only a retail quirk.
What actually helps
Pre-committed exits. A falsification condition written before entry. A time stop. A size cap that forces a trim when a winner has grown too large, so the “let it run” impulse has a boundary. A journal that records whether the sale matched the plan, so the pattern becomes visible in your own data rather than remaining a story you tell yourself.
Telling yourself to cut losers and let winners run does very little. Making the opposite behaviour slightly harder, and the planned behaviour slightly easier, does more.
What to do with this
Pull your last ten sales. Count how many were winners and how many were losers. If the winners dominate and the losers are still on the book, you are looking at the disposition effect in your own account. That is the cheapest research you will ever do.