Why you sell your winners and cling to your losers

Kahneman and Tversky built their theory on a very simple kind of experiment. A group is offered a choice between a certain gain and a gamble with the same expected value. The overwhelming majority takes the certain thing: they prefer certainty over risking a larger gain.

They are then given the same structure, but in losses: accept a certain loss, or gamble with a chance of losing nothing and a risk of losing much more. Here the majority flips completely: they reject the certain loss and prefer to gamble.

The same person is prudent when winning and a gambler when losing. This is not individual inconsistency: it is the general pattern, measured again and again since 1979. And it describes with unsettling precision what a trader does with a position in the red.

From that observation came prospect theory, the model of how people actually decide under risk — as opposed to how they ought to decide according to classical economic theory. Kahneman received the Prize in Economic Sciences in Memory of Alfred Nobel in 2002 for this line of work; Tversky had died years earlier.

Put the three together and you have a machine perfectly calibrated to destroy an account. The reference point tells you that you are losing. Loss aversion makes closing hurt twice as much as it would relieve. Distorted probabilities whisper that the comeback is likelier than it is.

The result of that combination has had a name since 1985, when Shefrin and Statman described it: the disposition effect, the tendency to sell winners too early and hold losers too long.

The mechanism is subtler than it looks and is worth following step by step. As long as you do not sell a losing position, the loss is only on paper and the mental account stays open. Selling closes it in red, and that act is what triggers the disproportionate pain. So you do not sell. With winners the opposite happens: closing in green produces immediate pleasure, so you close quickly — often too quickly.

Notice how perverse the design is. You are not deciding about the asset's future, which is the only thing that matters. You are deciding about the pain of admitting a mistake. The position is not the object of the decision: it is the excuse.

The consequences are measurable and all point the same way. Holding losers concentrates portfolio risk precisely in what is going wrong. Selling winners cuts off the right tail of the distribution, which is where most of the long-run return comes from. And in many jurisdictions it is also tax-inefficient, since realising losses can carry a tax benefit and realising gains the opposite.

Here the report takes a turn that usually surprises people. If these errors were random — each person erring in a different direction — they would cancel out and leave no trace in prices.

But they are not random. They are shared biases: nearly everyone errs in the same direction and at roughly the same time. And a systematic, synchronised error made by millions of people does not cancel out: it moves the price.

That is the theoretical reason persistent gaps between price and value exist. Not because the market is stupid, but because it is made of human beings who err predictably. Whoever understands the mechanism stops being the raw material of that gap and starts being able to study it.

It is worth being honest about the limit: understanding the bias does not immunise you against it. Studies document the disposition effect in professional managers too. Knowing it exists is necessary and is not sufficient.

The usual advice — “be more disciplined”, “control your emotions” — is precisely the advice that does not work, and the literature has spent decades showing why: it asks you to fight, in real time and under pressure, a bias that operates below conscious deliberation. It is asking someone to arm-wrestle their own nervous system.

What does work is removing the decision from the moment the bias is active. Three concrete moves, all of them before the pain:

1. Change the question. Not “how much am I down?”, which anchors on your purchase price, but “would I open this position today, at this price, knowing what I know now?”. If the answer is no, the position is not held up by analysis: it is held up by pride.
2. Reset the reference point. Your entry price carries no information about the asset's future. The market neither knows it nor cares. Every day is a fresh decision about the current price.
3. Write the rule before you hold the position. A decision made cold, in writing, and executed by a procedure rather than by your mood. The rule does not have to be perfect: it has to exist before the emotion does.

The third point is the central idea of this whole report. A bias that acts at the moment of deciding can only be beaten by a decision made before that moment. Everything else is willpower, and willpower has already lost this fight in every study where it has been measured.

All of the above has a direct consequence for how an analysis tool should be designed, and it explains a product decision that sometimes surprises people: there is no button here that says buy, and no score from one to ten.

A simple verdict is comfortable, and for that very reason dangerous: it becomes a new emotional reference point to cling to when the price goes against you. What disarms a bias is not one more opinion — it is a figure with its formula visible, its source data auditable and its date attached, so that you can honestly answer the only question that matters: would I open this today?

That is the role that belongs to us, and it is exactly the one a research desk performs inside an institution: producing the evidence somebody else decides on. The analyst does not place the order; they deliver traced data and a quantified scenario, and the decision passes through risk control before capital is committed. We do not issue the judgement. We produce the material it is issued from.