The Disposition Effect: Why Investors Sell Winners Too Early

Key Takeaway

A $100 Billion Mistake

In aggregate, the disposition effect is not a minor quirk. Odean (1998) estimated that the winners sold by individual investors outperformed the losers they held by 3.4 percentage points over the following year. Applied to the roughly $3 trillion in U.S. retail-held equities in the 1990s, the implied annual wealth destruction from premature winner-selling exceeded $100 billion. The figure has almost certainly grown with the expansion of retail trading. Barber and Odean (2000) subsequently documented that the most active individual traders underperformed a buy-and-hold strategy by 6.5 percentage points annually, with the disposition effect representing one of the largest identifiable components of that shortfall. This is not a rounding error. It is a behavioral tax that compounds relentlessly -- and unlike brokerage commissions, it has proven remarkably resistant to the forces of competition and technology.

The disposition effect is one of the most robust and costly behavioral biases in investing. Individual investors are roughly 50 percent more likely to sell a stock trading at a gain than one trading at a loss. This asymmetric behavior, rooted in prospect theory and loss aversion, drags portfolio returns by an estimated 3 to 4 percentage points annually. It also contributes to the momentum anomaly by delaying price discovery. Tax-loss harvesting -- selling losers rather than winners -- offers a rational and tax-efficient counter-strategy.

What Is the Disposition Effect?

In 1985, Hersh Shefrin and Meir Statman coined the term "disposition effect" to describe a pattern they observed in investor behavior: a strong tendency to sell assets that have appreciated in value (winners) while holding on to assets that have declined (losers). This runs directly counter to what rational tax optimization would suggest. Rational investors should defer gains to delay taxes and realize losses to harvest tax benefits.

The disposition effect is not confined to any single market or investor type. It has been documented among individual investors in the United States, Finland, China, Korea, and Australia. It appears in equity, options, and real estate markets. While professional investors exhibit the bias less strongly than retail traders, they are not immune.

The Theoretical Foundation: Prospect Theory

The most compelling explanation for the disposition effect comes from Kahneman and Tversky's prospect theory (1979), which describes how people evaluate gains and losses relative to a reference point -- typically the purchase price.

Prospect theory has three key features relevant to the disposition effect.

Empirical Evidence: The Odean Study

The landmark empirical study came from Terrance Odean in 1998. Using trading records from approximately 10,000 accounts at a large U.S. discount brokerage from 1987 to 1993, Odean calculated the Proportion of Gains Realized (PGR) versus the Proportion of Losses Realized (PLR).

Investors were 1.51 times more likely to sell a winning position than a losing one. This result held after controlling for portfolio rebalancing, tax-motivated trading, and the mechanical effect of limit orders.

Critically, Odean also examined subsequent performance. The winners that investors sold went on to outperform the losers they held by an average of 3.4 percentage points over the following 12 months. Investors were not cleverly selling stocks that had peaked. They were systematically cutting their best positions and riding their worst.

The Disposition Effect Across Markets

Subsequent research has confirmed the disposition effect globally, with interesting variations.

The bias is strongest among novice retail traders, in markets with high individual participation, and during periods of elevated volatility. Experience and sophistication reduce the effect but rarely eliminate it entirely.

How the Disposition Effect Fuels Momentum

One of the most important insights from behavioral finance research is that the disposition effect may be a key driver of the momentum anomaly.

Grinblatt and Han (2005) proposed a model in which the disposition effect creates a wedge between fundamental value and market price. When good news arrives, investors who are sitting on gains sell too quickly, creating excess supply and preventing the stock from reaching its fundamental value. When bad news arrives, investors holding at a loss refuse to sell, creating a floor that prevents the stock from falling to its fundamental value.

The result is systematic underreaction. Prices adjust slowly to new information because disposition-driven trading works against full price discovery. This creates the trending behavior that momentum strategies exploit.

This link also explains a puzzle in momentum research. Momentum is stronger among stocks with high individual ownership and weaker among stocks held primarily by institutions. If the disposition effect drives momentum, this is exactly what we would expect -- the bias is stronger among retail investors.

Tax-Loss Harvesting: The Rational Counter-Strategy

If the disposition effect causes investors to hold losers and sell winners, the optimal strategy does the opposite. Tax-loss harvesting involves systematically selling positions at a loss to realize capital losses, which can be used to offset capital gains and reduce tax liability.

The mechanics are straightforward.

Step 1. Periodically screen the portfolio for positions trading below their cost basis by a meaningful threshold, typically 5 to 10 percent.

Step 2. Sell the losing position and immediately reinvest in a similar but not identical asset to maintain market exposure. In the U.S., the wash-sale rule prohibits repurchasing substantially identical securities within 30 days.

Step 3. Book the capital loss and use it to offset realized gains, or up to 3,000 dollars of ordinary income per year in the U.S., carrying forward unused losses to future years.

Wealthfront and Betterment have published studies suggesting that systematic daily tax-loss harvesting can add 1.0 to 2.0 percent per year in after-tax returns for taxable accounts, depending on portfolio volatility and the investor's tax bracket.

The behavioral insight is critical: tax-loss harvesting forces you to do exactly what your instincts resist -- sell losers and redeploy capital into positions with better forward-looking prospects.

Practical Implications for Quantitative Investors

For systematic strategies. The disposition effect is a pure alpha source for momentum-based strategies. If other market participants systematically underreact because they sell winners too early and hold losers too long, price trends will persist longer than they otherwise would. Momentum strategies profit from this.

For portfolio construction. Quantitative investors should build rules that counteract the disposition effect. Automated rebalancing and systematic stop-losses eliminate the human tendency to anchor on purchase prices. Position sizing should be based on forward-looking signals, not on whether a position is currently profitable.

For risk management. The disposition effect creates hidden portfolio risk. Investors who refuse to sell losers end up with concentrated positions in declining stocks. This increases portfolio volatility, reduces diversification, and can lead to catastrophic drawdowns.

Applied Analysis: The Disposition Effect Across Investor Types and Market Conditions

The magnitude of the disposition effect varies systematically with investor characteristics, market conditions, and institutional structure. The following table compiles findings from major empirical studies to illustrate these patterns.

Several patterns emerge. First, the disposition effect is remarkably universal -- it appears across every investor type and market studied, with variation in magnitude but not direction. Second, the effect is strongest among the least sophisticated investors (retail traders in high-turnover markets like China) and weakest among professionals with institutional constraints. Third, market conditions modulate the effect: bull markets amplify the bias because there are more winning positions to sell prematurely, while bear markets compress it mechanically.

The real estate evidence is particularly striking. Genesove and Mayer (2001) documented that homeowners selling at a nominal loss set asking prices approximately 25-35 percent above market value compared to those selling at a gain, dramatically extending time-on-market and often resulting in final sale prices below what earlier acceptance would have achieved. The disposition effect operates even in markets with infrequent transactions and high information costs.

Competing Frameworks: Disposition Effect vs. Alternative Explanations

The disposition effect is the dominant behavioral explanation for asymmetric holding behavior, but alternative theories offer competing or complementary explanations. Understanding where these frameworks converge and diverge is essential for designing effective de-biasing strategies.

Prospect Theory / Disposition Effect. Shefrin and Statman (1985) grounded the disposition effect in Kahneman and Tversky's prospect theory: the S-shaped value function makes investors risk-averse in the domain of gains (selling winners early to lock in certain gains) and risk-seeking in the domain of losses (holding losers in hope of recovery). The reference point -- typically the purchase price -- determines whether a position is coded as a gain or loss. Prediction: investors sell winners and hold losers regardless of fundamentals.

Mean Reversion Belief. An alternative rational explanation posits that investors sell winners and hold losers because they believe in mean reversion -- stocks that have risen are "due" to fall, and stocks that have fallen are "due" to recover. Odean (1998) tested this directly and found that the winners sold actually continued to outperform the losers held, ruling out correct mean-reversion beliefs as an explanation.

Overconfidence. Barber and Odean (2001) documented that overconfident investors trade more frequently and incur higher costs. Overconfidence can interact with the disposition effect: overconfident investors may sell winners to redeploy capital into new "better" ideas, while refusing to sell losers because admitting the loss would challenge their self-image as skilled traders. The two biases are synergistic rather than competing.

Regret Aversion. Bell (1982) and Loomes and Sugden (1982) proposed that investors anticipate the regret they would feel from different outcomes. Selling a winner that continues to rise produces regret, but less than the regret from realizing a loss. This creates an asymmetric reluctance to sell losers. Regret aversion predicts similar behavior to prospect theory but through a different psychological mechanism.

The weight of evidence favors prospect theory as the primary driver, with overconfidence and regret aversion as amplifying mechanisms. Mean reversion belief is the only framework that has been empirically rejected as a standalone explanation.

Reassessing the Disposition Effect

The disposition effect is well-documented, but several nuances deserve attention. The estimated magnitude of the return drag (3 to 4 percentage points) comes primarily from retail brokerage data from the 1990s. Commission-free trading, automated investing, and greater financial literacy may have reduced the effect among certain investor segments, though evidence from Robinhood-era studies suggests the bias remains strong among new cohorts of retail traders. The causal link between the disposition effect and momentum profitability, while theoretically elegant, is debated -- other behavioral and structural factors also contribute to momentum. Tax-loss harvesting benefits depend heavily on jurisdiction-specific tax rules and are less relevant for tax-exempt accounts. Finally, the disposition effect interacts with other biases, including the endowment effect and status quo bias, making it difficult to isolate in complex portfolio contexts.

What the Trading Data Shows

The disposition effect holds a distinctive position in behavioral finance: it is simultaneously one of the oldest documented biases and one of the most empirically robust. The core finding -- that investors are approximately 50 percent more likely to sell winning positions than losing ones -- has been replicated across more studies, markets, and time periods than almost any other behavioral phenomenon in finance.

Replication strength: Exceptionally strong. Odean's (1998) original PGR/PLR ratio of 1.51 has been confirmed by Grinblatt and Keloharju (2001) in Finland, Feng and Seasholes (2005) in China, Shapira and Venezia (2001) in Israel, and numerous other studies across Asia, Europe, and Australia. The effect has been documented in equities, options, real estate, mutual funds, and even cryptocurrency markets. No major replication has failed to find the effect, though its magnitude varies across contexts.

Causal mechanism evidence. The prospect theory explanation is supported by Barberis and Xiong (2009), who showed that a model combining annual gain-loss utility evaluation with prospect theory preferences closely replicates the observed selling patterns. Frazzini (2006) provided institutional evidence that mutual fund managers exhibit the disposition effect and that it creates predictable price patterns exploitable by momentum strategies. Weber and Camerer (1998) demonstrated the disposition effect in controlled laboratory experiments, ruling out market microstructure as the primary cause.

Magnitude and economic significance. Odean (1998) estimated a 3.4 percentage point annual cost. Barber and Odean (2000) found that the most active traders underperformed by 6.5 percentage points annually. Dhar and Zhu (2006) showed that wealthier and more financially literate investors exhibit a weaker disposition effect but do not eliminate it entirely. As of 2025, the consensus estimate of the annual return drag from the disposition effect for typical retail investors ranges from 1.5 to 4.0 percentage points, with the range reflecting variation in trading frequency, portfolio size, and market conditions.