Behavioral FinanceTrading Psychology

The Disposition Effect: Why Investors Sell Winners Too Early and Hold Losers Too Long

Summary by Robert Gorak · Published June 11, 2026 · Last reviewed June 18, 2026

Hersh Shefrin and Meir Statman·1985·Journal of Finance
Sample: 252 months of mutual fund aggregate data (Broker/Dealer and Direct Seller); 156 months (No-Load); individual investor trades from Schlarbaum et al. (1964–1970)Data: Schlarbaum, Lewellen, and Lease individual investor panel; Investment Company Institute mutual fund dataPeriod: Individual investors: 1964–1970; Mutual funds (Broker/Dealer, Direct Seller): January 1961–December 1981; No-Load funds: January 1961–December 1973

The disposition effect describes investors' tendency to sell winning positions too early while holding losing positions too long. An investor who holds a stock down from $50 to $40 often refuses to sell, waiting to break even instead. Shefrin and Statman (1985) found redemption ratios of 0.93 in gain months versus 0.74 in loss months across 252 months of Broker/Dealer data (1961–1981). Their paper, "The Disposition to Sell Winners Too Early and Ride Losers Too Long," identifies four drivers: prospect theory, mental accounting, regret aversion, and self-control.

What the Study Found

In the Schlarbaum et al. individual investor data (1964–1970), approximately 40% of all realizations were losses across every round-trip duration category. The gain-realization ratio was 0.58 for 0–1 month holds, 0.57 for 2–6 months, and 0.59 for 7–12 months. In Broker/Dealer mutual funds (month t−1 specification), the mean redemption ratio was 0.93 in capital-gains months versus 0.74 in capital-losses months. The t-statistic was 1.69, significant at the 0.05 level — the only significant result across all 9 fund-type and timing combinations tested. No-Load and Direct-Seller funds showed directionally similar but insignificant differences (t-statistics ranging from −0.09 to 1.62).

Methodology

The individual investor stock data came from Schlarbaum, Lewellen, and Lease, covering brokerage panel trades from 1964 to 1970. Mutual fund data came from the Investment Company Institute: monthly purchases and redemptions for three fund types, January 1961–December 1981 (252 months). No-Load fund data ended in December 1973; post-1973 money market funds used accrual accounting, making loss realization structurally impossible. The study partitioned individual trades by round-trip duration (0–1, 2–6, 7–12 months) and compared redemption ratios across 30 highest-gain and 30 highest-loss months.

Key Statistics

Metric Finding Context
Gain-realization ratio, 0–1 month round-trips 0.58 Schlarbaum et al. individual investor data, 1964–1970
Gain-realization ratio, 2–6 month round-trips 0.57 Schlarbaum et al. individual investor data, 1964–1970
Gain-realization ratio, 7–12 month round-trips 0.59 Schlarbaum et al. individual investor data, 1964–1970
Mean redemption ratio, Broker/Dealer funds, capital-gains months (month t−1) 0.93 30 highest-gain months, January 1961–December 1981
Mean redemption ratio, Broker/Dealer funds, capital-losses months (month t−1) 0.74 30 highest-loss months; t = 1.69, p < 0.05
Mean redemption ratio, Direct-Seller funds, capital-gains months (month t−1) 0.92 30 highest-gain months, January 1961–December 1981
Mean redemption ratio, Direct-Seller funds, capital-losses months (month t−1) 0.66 30 highest-loss months; t = 1.62 (not significant)
t-test for redemption ratio difference t = (X̄_G − X̄_L) / sqrt((N_G·S_G² + N_L·S_L²) / (N_G + N_L − 2)) · sqrt(N_G·N_L / (N_G + N_L)) N_G = N_L = 30; tests H₀: X̄_G = X̄_L against H₁: X̄_G > X̄_L (Table II)

Why This Matters

The disposition effect challenges purely tax-based models of investor behavior. Even when realizing a loss would be optimal under Constantinides' normative strategy, investors systematically avoid it. The framework integrating prospect theory, mental accounting, regret aversion, and self-control offers a richer explanation of observed trading patterns than rational models alone. December loss concentrations, the paper argues, reflect self-control deadlines rather than rational optimization, with implications for tax-loss harvesting and year-end return seasonality.

Frequently Asked Questions

0.58, 0.57, and 0.59 — the gain-realization ratios for 0–1, 2–6, and 7–12 month holds in the Schlarbaum et al. data — are nearly identical. Tax-motivated predictions require a very low ratio for short-term holds and a high ratio for 7–12 month holds. The flat empirical pattern is inconsistent with tax optimization as the sole driver.

1 of 9 fund-type and timing combinations showed a statistically significant difference in redemption ratios. In Broker/Dealer funds, the mean ratio was 0.93 in capital-gains months versus 0.74 in capital-losses months (t = 1.69, p < 0.05). No-Load and Direct-Seller fund results were directionally consistent but did not reach statistical significance.

Shefrin and Statman argue December serves as a perceived self-control deadline for loss realization, not a rational tax-planning date. A rational investor could forecast tax liability at year-start and adjust withholding to realize the benefit immediately. The paper shows December concentration is inconsistent with Constantinides' rational model but consistent with their behavioral framework.

Four elements form the paper's theoretical basis. Prospect theory's S-shaped value function creates risk aversion for gains and risk seeking for losses. Mental accounting tracks gains and losses from a reference purchase price. Regret aversion adds reluctance to close losing accounts. Self-control — a planner-doer conflict — is fourth; tax considerations are fifth.

Source

Hersh Shefrin and Meir Statman (1985). The Disposition to Sell Winners Too Early and Ride Losers Too Long: Theory and Evidence. Journal of Finance.

Read the full paper