Why you sell winners and hold losers

What it is

Investors tend to sell positions that are up and keep positions that are down — far more than a tax or portfolio logic would explain. The pattern was named by Shefrin and Statman in 1985, building on earlier observations that accounts showed a tendency to “sell the winners and ride the losers.” The striking part is not that it happens; it is how consistently it shows up across markets and decades.

What the evidence shows

• US brokerage accounts (Odean, 1998). Across roughly 10,000 accounts from 1987–1993, a stock that had risen was about 60% more likely to be sold than one that had fallen (outside December). Realized gains averaged about 14.8%; realized losses about 9.8%. In December the pattern flipped — consistent with tax-loss selling, which is a reminder that the effect is context dependent.
• Taiwan (Barber et al.). Investors sold winners roughly twice as often as losers, and about 85% of the sample showed the pattern.
• Finland (Grinblatt & Keloharju; Seru et al.). Investors were about 36% more likely to sell extreme losers in December, and the behavior has been estimated to cost the average investor somewhere between 3.2% and 5.7% in returns.
• Futures traders (Locke & Mann). Disposition-prone futures traders took on more risk and earned less — the effect is not confined to retail stock accounts.
• Professionals are not immune. Mutual fund managers show the pattern to a lesser degree; underperforming funds were about 1.7× more likely to realize gains than losses (Frazzini).

Why it happens

Shefrin and Statman attributed it to several forces at once: loss aversion as described by prospect theory (losses feel roughly twice as heavy as equal gains — Tversky & Kahneman, 1979), mental accounting tied to the purchase price, regret avoidance, and self-control conflicts. Later work (Ben-David & Hirshleifer) argues that belief updating may matter more than a direct preference for realizing gains. The honest summary: nobody has a single complete explanation, and no single element — including loss aversion — accounts for the whole effect.

What experiments say can interrupt it

Weber and Camerer's experiments reproduced the pattern in the lab and found that automatically selling a fraction of each position every period — a rule fixed in advance — greatly reduced it. That is a statement about rules, not about markets: a decision made in advance, when nothing is at stake, appears to hold better than a decision made in the moment. It is also the one mechanism this product actually implements (a commitment made before the day starts, settled after it ends).

What is still disputed

• Whether the driver is a preference (realizing gains feels good) or belief updating.
• Whether the return cost estimates generalize across markets and eras.
• How much of the pattern is tax-driven — December behavior suggests a meaningful share.
• Whether professional investors simply learn to suppress it, or face different incentives.

What this product does — and does not do

We do not claim to fix the disposition effect, and we cannot see your positions. What we offer is the one lever the experiments point at: a rule made in advance, plus a record of whether it held. The state check-in exists so that the state you are in — not the market — is the first thing you look at.

Nothing on this page is financial advice, and nothing here predicts any market.