Three checkpoints in this guide
Follow the full walkthrough in order, or jump directly to one of its main sections.
Selling a winner can feel safe: the gain is now real. Closing a loser can feel like admitting that the original idea failed. Researchers call the tendency to realize gains more readily than losses the disposition effect. But a winning trade held for ten minutes and a losing trade held for an hour do not, by themselves, diagnose it. Different strategies intentionally have different exit paths. The useful question is whether your exits departed from a rule you could have stated before seeing the result—and whether that departure hurt after costs and risk.
The short answer
In a study of 10,000 US discount-brokerage accounts, Terrance Odean found investors were more inclined to realize gains than losses. In the paper's full-year test, the proportion of gains realized was 0.148 versus 0.098 for losses. These are fractions of available gain and loss opportunities under a particular stock-portfolio method—not the percentage of all customers who made money, not a win rate and not a probability that any new trade should be sold. A Korean stock-index futures study also found a disposition effect, especially among individual investors, and reported a negative relationship with performance. Yet a separate study of professional floor futures traders found they held losses longer than gains without finding measurable costs from that behavior in its sample. Odean's original paper; Korean futures study; professional futures study.
The evidence therefore supports a question, not a slogan: is your winner/loser exit asymmetry a tested part of the strategy, or a repeated override made after P/L becomes emotionally salient?
What the studies can—and cannot—establish
| Primary research | What the authors observed | Boundary for your trade review |
|---|---|---|
| Odean, Are Investors Reluctant to Realize Their Losses?, US discount-brokerage stock accounts, 1987–1993 | Investors realized a higher fraction of available gains than available losses; the full-year rates were 0.148 and 0.098 under the paper's definitions. | An equity portfolio with daily prices and unsold positions is not a 2026 futures or crypto account. The fractions are not a personal diagnostic threshold. |
| Choe and Eom, The Disposition Effect and Investment Performance in the Futures Market, Korean stock-index futures | Individual investors showed a stronger disposition effect than institutional/foreign investors; the measured effect had a negative relationship with performance. | This market-specific association does not prove that copying one exit rule would increase another trader's profits. |
| Locke and Mann, Professional Trader Discipline and Trade Disposition, professional floor futures traders | Traders held losses significantly longer than gains, but the authors found no associated measurable cost in their sample; successful traders also showed disciplined behavior. | Holding time alone is not a verdict. Professional floor inventory and information conditions differ from retail screen trading. |
Citable interpretation: Realizing gains more readily than losses is documented in some investor and futures datasets. It is not a universal rule that every longer-held loser is irrational or costly. The studies differ in participants, trading venue and definition of an opportunity; no pooled “X% of traders do this” figure is warranted. This is TSB's synthesis of the cited research, not a study of TSB users.
Why the denominator matters more than a catchy ratio
Odean did not simply count how many sold stocks were winners. A rising market can produce more winning holdings than losing ones, so a neutral investor may sell more winners just because more are available. The paper compared realized gains with both realized and paper gains, and realized losses with both realized and paper losses, on the relevant sale days. It also examined a December period when tax incentives alter the pattern. Original method and results.
A closed-trades CSV lacks the unsold positions and contemporaneous prices needed for that same opportunity denominator. If all you know is that seven winning trades closed quickly and three losing trades closed slowly, you can describe an exit-duration pattern. You cannot state that the trader was more likely to realize every available gain than every available loss, or that psychology caused the difference. A strategy may deliberately take quick mean-reversion gains and give a well-defined invalidation more time. Conversely, a trend-following plan may require keeping some winners open longer; cutting them at the first green number would be a meaningful deviation even if the final week stayed profitable.
Three different explanations that can look the same
A legitimate payoff design. A short-target strategy can take many small gains and accept fewer, larger planned losses if net expectancy and tail risk still pass its predeclared test. Longer loser duration is not automatically evidence of denial. The win-rate and risk-reward guide owns the arithmetic of that payoff shape.
An execution or data artifact. Partial exits, a trailing stop, funding or overnight fees, differences between opened and closed legs, and timezone mismatches can change both apparent holding time and net P/L. A single “trade” may be several fills, while a closed-only file omits the path taken before exit. First reconcile the trade grain and costs.
An after-entry rule change. A trader may move a losing stop away from the original invalidation, add size without a predeclared plan, or take profit early because the open gain feels uncomfortable. That is a testable plan-versus-execution mismatch if the original plan and modification are recorded. A profitable deviation is still a deviation; one losing trade within plan is not proof of bad discipline.
The studies do not let us read a person's motive from P/L. They make the asymmetry worth investigating, not diagnosing by intuition.
A worked trade review: observation versus accusation
Suppose a trader's plan sets an initial risk of $100 and a target of $200 for a named setup. In a hypothetical session, one entry closes at +$65 after an unplanned manual take-profit. Another reaches the original stop at −$100. The combined result is −$35 before fees. Calling this “proof of the disposition effect” would be too strong: it is only two trades, and the missed target might never have been reached. Calling the first exit “good because it won” would be too weak: it changed the payoff rule that the strategy was supposed to test.
The defensible record is: one early winner was a rule deviation, one loser followed plan, the small sample lost money, and the alternative exit result is unknown without a price-path and execution simulation. A later set of comparable trades can test whether early manual profit-taking changes net expectancy or downside. The MAE/MFE guide explains how intratrade excursion data can generate exit hypotheses without pretending the best observed price was always executable.
How to test your own exits without inventing evidence
- Freeze the intended rule. Record the setup, initial stop, target or trailing method, risk amount, time limit and conditions for a discretionary exit before the entry. Version changes rather than retroactively editing the old plan.
- Reconcile each decision unit. Pair partial fills and scale-outs, choose a consistent timezone, include known costs and keep deposits or account transfers out of trade P/L. Identify open positions separately from closed trades.
- Label the exit reason, not the emotion you imagine. Distinguish target, stop, trailing rule, time exit, planned scale-out, manual override, operational failure and unknown. “Fear” requires the trader's own contemporaneous note; a short winner alone does not establish it.
- Compare like with like. Within each stable setup and market regime, show the count of eligible opportunities, winner/loser holding times, planned versus actual R, net outcomes, largest losses, and missing plans. If open-position snapshots and prices are available, assess unrealized opportunities separately; do not substitute a closed-trade win rate for them.
- Predeclare one prospective change. For a suspected early-profit pattern, test a specific target/exit variant on later comparable trades, with realistic fills, fees, stops and drawdown. Keep skipped and failed trades in the denominator. A historically selected best exit is a hypothesis, not proof that the trader should have held every winner longer.
The answer may be “not verified”. If the plan, intratrade path or complete costs are absent, describe the observable exit pattern and the missing evidence. There is no research-backed rule to “always let winners run”; an unbounded loss is not the price of avoiding the disposition effect.
Where TSB fits
TSB can help organize selected, recorded trade history by setup, session, account, exit outcome and available costs, so the trader can ask its AI Coach which comparable trades support a suspected pattern. The Coach's answer should identify its scope and missing data. A closed-trade import cannot reconstruct every unsold paper gain or loss in Odean's portfolio test, and a screenshot taken after exit cannot prove the original stop or emotion. The trader's recorded plan and a verified price path are separate inputs when available. AI Coach access also depends on the current product entitlement; the free journal alone must not be presented as a promise of unrestricted AI analysis.
The productive next question is specific: “Among trades where my original plan and actual exit are both recorded, did early manual profit-taking reduce after-cost results on later comparable setups?” Start with your own recorded trading history rather than a rule inferred from strangers' historical accounts.
Bottom line
The disposition effect is a real research finding in some historical stock and futures samples. The professional-futures counterexample shows why holding losers longer than winners is not, by itself, a diagnosis or a quantified cost. First separate strategy design from unplanned overrides; then reconcile trades, include the opportunities your data actually covers, and test one exit change on later comparable evidence. That turns a familiar slogan into a decision you can audit.
Source and scope note
The cited US stock portfolios, Korean stock-index futures accounts and professional floor futures traders are different populations. We do not average their results, attribute them to TSB customers, infer a universal 2026 prevalence, or claim a behavioral cause for a particular trader. For method and corrections, see TSB's editorial policy.
