Why do investors sell winners and hold losers?

The short answer

The pattern is one of the most robust findings in retail investor behavior. Across 10,000 discount brokerage accounts spanning 1987-1993, Terrance Odean (1998) found that investors realized gains roughly 50% more frequently than they realized losses, even after controlling for portfolio composition and tax considerations.

What makes this finding remarkable is the subsequent return data. Stocks that investors sold went on to outperform the stocks they kept by approximately 3.4 percentage points per year over the following 12 months. The behavior is not just psychologically driven — it is economically costly in a measurable way.

The pattern intensifies in environments where trading frictions are low and account checking is frequent. Mobile trading apps, real-time portfolio dashboards, and zero-commission brokerages have all been associated with measurable increases in disposition-effect trading.

→ New to behavioral finance? Investing for Beginners Hub

What the data shows

The empirical record on sell-winners-hold-losers behavior spans four decades and crosses multiple jurisdictions and asset classes.

The numerical context (Odean 1998, Shefrin-Statman 1985, brokerage data 1987-2024) :

• Proportion of gains realized vs available: ~14.8% (Odean 1998)
• Proportion of losses realized vs available: ~9.8% (Odean 1998) — gains realized ~50% more frequently
• Subsequent 12-month return spread: stocks sold beat stocks held by ~3.4 pp/year
• Effect persists in Israeli, Finnish, Chinese and Taiwanese retail data with similar magnitude
• Mobile-app users show ~30% higher disposition-effect coefficients than desktop-only traders (academic literature 2018-2022)

The exception : at year-end, the pattern partially reverses as tax-loss harvesting incentives temporarily push investors to realize losses. The reversal is incomplete and concentrated in December trading days, suggesting it is rule-driven rather than evidence of overcoming the underlying bias.

→ Dataset: S&P 500 Historical Returns

Why it happens — the macro mechanism

Three mechanisms combine to produce sell-winners-hold-losers behavior at scale.

Realization-utility channel. Investors derive utility from the act of realizing a gain — a positive emotional event independent of total wealth. Conversely, realizing a loss generates a negative emotional event. The asymmetry biases the timing of trades toward locking in gains and deferring losses, even when after-tax expected returns favor the opposite trade.

Mean-reversion belief channel. Many retail investors implicitly hold a mean-reversion model: stocks that fell will rise, stocks that rose will fall. This belief is partially correct over very long horizons but systematically wrong over the 6-12 month horizons most relevant to portfolio decisions. Recent winners tend to exhibit modest momentum, while recent losers often continue underperforming — directly inverting the implicit mean-reversion prior.

This second channel connects directly to the dynamics of momentum factor returns.

Reference-point dependence channel. Each holding has its own purchase price, anchoring perceived gain or loss in isolation. This atomization makes investors evaluate positions individually rather than at the portfolio level, missing the basic insight that what matters is forward expected return, not historical entry price. The momentum factor systematically extracts value from this bias.

Synthesis by regime : in trending bull markets (2013-2019, 2020-2021), the disposition effect produces severe underperformance because winners continue winning and losers continue losing; in range-bound markets (2015-2016), the bias has less measurable cost because mean reversion partially validates the implicit belief; in trending bear markets (2008, 2022), the cost is asymmetric — held losers compound losses while winners that were sold have already left the portfolio. The regime that hurts disposition-affected investors most is sustained directional trending, particularly when accompanied by high cross-sectional dispersion.

The disposition effect doesn’t just reflect bias — it transfers ~3 percentage points per year from those who feel to those who measure.

→ Framework: Behavioral investing pillar

What it means for different economic actors

Retail investors. The disposition effect is most measurable in this group. Unconstrained by mandate or career risk, retail investors display the rawest form of the bias — and pay the largest cost in foregone returns.

Active fund managers. Disposition-effect coefficients are substantially smaller for institutional managers, but career risk substitutes a related distortion: holding underperformers while quietly trimming outperformers to lock in benchmark-relative gains.

Quantitative funds. Systematic strategies often explicitly exploit the disposition effect via momentum factors, capturing the spread between sold winners and held losers as a structural source of returns. Loss aversion drives the bias; momentum captures it.

A common error is to view the bias as a tax-optimization problem. While taxes amplify the underperformance, even tax-exempt accounts (IRAs, pension trusts) display the same pattern — the root cause is psychological, not fiscal.

Practical observation

Go deeper

Related questions

• What is the disposition effect in practice?
• What is loss aversion and how does it affect investment decisions?
• What is the momentum factor in markets?
• Why is overconfidence the most expensive investor bias?

Frequently asked questions

Is selling winners always a mistake?

No. Selling winners makes sense when forward expected returns are deteriorating, when concentration risk has grown excessive, or when rebalancing rules require it. The bias is not selling winners per se but selling them preferentially over losers regardless of forward expected return. Disciplined sellers who apply the same evaluation criteria to all positions show no disposition effect — and no measurable underperformance.

Why does the 3.4 pp/year cost persist if it is well known?

Because the bias is psychological, not informational. Investors who read about the disposition effect typically continue to display it in their own trading. Knowing about a cognitive bias is a much weaker remedy than engineering around it through structural rules — automatic rebalancing schedules, pre-committed sell criteria, or systematic strategies that ignore entry prices entirely.

How does this connect to momentum factor returns?

Quite directly. The momentum factor — buying recent winners and shorting recent losers — captures roughly the inverse of disposition-effect trading. Academic decompositions suggest the disposition effect contributes meaningfully to momentum’s positive risk premium, particularly in retail-heavy markets like Korean and Chinese equities. The factor is, in part, a structural transfer from disposition-affected traders to systematic ones.

Last updated — 22 May 2026