Why You Cut Winners Short and Let Losers Run

Take a survey of trading mistakes and this one wins every time: I cut my winners too early and hold my losers too long. Traders say it about themselves with a shrug, as though it were a character flaw like being bad at mornings.

It is not a character flaw. It is the most robustly documented bias in the whole of behavioural finance, it has a name — the disposition effect — and its cost has been measured on real accounts.

The measurement

In 1998 Terrance Odean published an analysis of trading records from 10,000 accounts at a US discount broker over 1987–1993. He asked a narrow question: when an investor sells something, how does the decision depend on whether the position is up or down?

The answer was unambiguous. Investors realised gains at a rate of 0.148 and losses at a rate of 0.098 — roughly one and a half times more willing to close a winner than a loser, with a t-statistic of 35. That is not a tendency. That is a law of the sample.

The one month it reversed was December, when tax-loss selling briefly made losses attractive to realise. Give people an external reason to book a loss and they book it. Remove the reason and they will not.

The part that costs money

Selling winners early is only a mistake if the winners were going to keep winning. Odean checked.

Over the following year, the winners investors sold beat the market by +2.35%. The losers they held on to trailed it by −1.06%. A gap of 3.41 percentage points, statistically significant at p = 0.001, and it persisted over two years.

They did not merely sell the wrong ones. They sold the wrong ones by a measurable, repeatable, annually-compounding margin — and every one of those decisions felt, in the moment, like prudence.

Why it happens, mechanically

The standard explanation runs through prospect theory: gains and losses are evaluated against a reference point, usually your entry price, and the value function is concave in gains and convex in losses. In plain language: the certainty of a small win is more attractive than the possibility of a larger one, and the possibility of getting back to flat is more attractive than the certainty of a loss.

Loss aversion sits underneath it — the finding that losses weigh more heavily than equivalent gains. The famous coefficient of 2.25 is worth treating carefully; it came from a small unincentivised experiment in 1992, and a 2024 meta-analysis of 607 estimates puts the true figure closer to 1.95. But the direction has never been in dispute. Roughly speaking, losing hurts about twice as much as winning feels good, which is exactly the asymmetry that makes a scratch feel like a victory and a stop-out feel like a verdict.

Notice what all of this is measured against: your entry price. That number is the source of the entire distortion, and it is the one input in the whole process that contains no information about the market. The chart does not know it. Every other participant is trading without it.

The trading-specific version

The Odean data is US household stock accounts, not futures, and the mechanism transfers with one important amplification: leverage puts the disposition effect on a clock. An equity investor holding a loser is having an unpleasant year. A futures trader holding a loser is having a margin conversation, and the position will be resolved with or without them.

The specific behaviours are easy to name, and most traders will recognise at least two:

• Taking a 1.2R exit on a trade that was structurally headed for 3R, because the profit was “real money now.”
• Moving the stop to breakeven the moment a trade is green — which, as the lab experiment found, removed 44% of a system's expectancy in exchange for a better-looking win column.
• Widening a stop on a loser while never widening a target on a winner. The asymmetry is the tell.
• Scratching a position out of boredom at +0.3R, then watching it complete the move you had correctly identified.

Each is individually defensible. Together they are a system for guaranteeing that your average winner is smaller than your average loser, which is the one condition under which a good method still loses money.

What actually works against it

Define the exit before the entry, in the market's language. “Out at the next level up” is a statement about the chart. “Out at +$400” is a statement about your entry price, and it will drift with your mood. The first can be defended while the trade is running; the second cannot.

Make the target a rule with a reason. If the plan is a 3R target and the trade reaches 1.5R, the only legitimate grounds for exiting early are that something on the chart changed — not that the number got big enough to be tempting. Write the reason in the journal at the moment of exit. Three weeks of doing that produces an honest inventory of how often the reason was “I wanted it.”

Measure your average winner against your average loser, monthly. This single ratio catches the disposition effect faster than any amount of self-observation. If your average win is 1.1R and your average loss is 1.0R while your plan says 3R and 1R, the leak is diagnosed and it is in the exits, not the entries.

Consider taking the reference point away from yourself. Partial exits at a pre-planned level, a trailing rule defined by structure, a bracket order placed at entry — anything that puts the decision into a mechanism executes it while you are calm. The expectancy calculator will show you exactly what a fractional R is worth over a hundred trades, which is generally more sobering than any argument.