OptiNod Academy
The disposition effect — Taking gains early while letting losses run
A high win rate can coexist with a shrinking account when exit habits are asymmetric. Average holding time and realized-gain versus realized-loss rates reveal the pattern.
When an account shrinks despite a high win rate, the problem is often at the exit. Closing gains quickly while holding losses can be measured in the trade record.
The disposition effect was named by Hersh Shefrin and Meir Statman in 1985. It is the tendency to realize winning positions quickly and keep losing ones longer. Both actions share a cause. Traders judge gain or loss against their entry price. In profit, they avoid risk for fear of losing what they have; in loss, they seek risk in the hope of getting back to even. Prospect theory's value function is at work in the exit decision.
The habit is often reduced to “not having the nerve to stop out.” If it is only a willpower problem, the remedy becomes “be tougher,” which can fail on the next loss. Because the disposition effect arises from an asymmetric experience of gains and losses, it is better addressed with rules that avoid making a fresh emotional exit decision.
Its cost is hidden from win-rate statistics. Frequent small realized gains keep win rate high, but average losses outgrow average wins and expectancy turns negative. That is how a trader can feel successful at a 60% win rate while the account declines. Even with 65% wins, an average gain of 1.0R and average loss of 2.0R yield expectancy of 0.65 × 1.0R − 0.35 × 2.0R = −0.05R per trade. A high win rate does not guarantee positive expectancy. Two measures in the record reveal the asymmetry.
The value function makes exits asymmetric
Prospect theory's value function explains the mechanism. A loss of a given size feels roughly twice as powerful as a gain of the same size, and the trader uses entry price as a reference point. Together they produce opposing exit habits.
In profit, risk aversion appears. The chance that a position with a $100 gain returns to zero feels more pressing than the chance it grows to $200, so the small certain profit is taken quickly. In loss, risk seeking takes over. Realizing a $100 loss feels painful, so the trader holds for a small chance of getting back to break-even. Both choices can feel sensible separately; together they create a distribution of short winners and long losers.
That distribution directly lowers reward relative to risk. A trend system uses a few large wins to cover many small losses. The disposition effect cuts those rare large wins short. Even if entry rules have positive expectancy, trimming the positive tail at the exit can leave live expectancy below the backtest.
Closing profits quickly can miss an entire large trend
Bitcoin rose from about $69,000 in early November 2024 to $108,353 on December 17, a gain of roughly 57%. A trader who entered early might gain 8% in a few days and close near $75,000 to “protect the profit.” The entry analysis and signal were right, yet price later rose more than another 40% from that exit.
No loss was booked, so the trade appears as a win in the win-rate count. The problem emerges over repeated trades. Habitually closing a large trend at +8% turns an occasional +57% winner into +8%, lowering average gains. Trend-following averages depend on rare large winners; frequent small wins contribute much less. The disposition effect removes the largest winners first.
The pattern repeated in 2025. Bitcoin rose about 67% from its April 7 low of $74,508 to $124,474 on August 14. A trade closed around +10% during that four-month rise was not a losing trade, but it surrendered much of the average gain that one large trend could have produced.
Holding losses can turn a planned 1R into many R
The other side is more costly. Bitcoin reached about $126,200 on October 6, 2025 and then fell around 36% to $80,600 by November 21. Suppose a buyer near the high planned a 5% stop, around $119,900. An exit there would have limited the loss to the planned 1R.
The loss-side disposition effect makes the trader postpone it. At $119,900, “wait a little longer for a rebound” replaces the stop. As price falls, the hope of break-even gets stronger. Holding to $80,600 produces about a 36% loss, more than seven times the planned 1R, or below −7R. One such trade can cancel the benefit of seven normal stops.
An early-2026 fall illustrated the same risk faster. Bitcoin was around $98,000 in mid-January and fell roughly 39% to $60,000 by February 6, in a little over three weeks. A postponed stop ended far deeper than planned. In the moment it may feel like “flexibility”; in the record it appears as one of a few outsized losses lifting the average loss.
Two measures reveal the disposition effect
Treating this only as willpower leaves both diagnosis and correction vague. The trade record provides two measurements.
First is average holding time. Divide trades into winners and losers and average each group's duration. With a disposition effect, winners are held for less time than losers. A gap such as three days for winners and 19 days for losers is a clear sign that profits are closed early while losses linger.
Second is the realization rate. Terrance Odean proposed PGR and PLR while analyzing roughly 10,000 individual accounts in 1998.
- PGR, proportion of gains realized: Number of realized winning positions divided by realized winning positions plus currently held positions with gains.
- PLR, proportion of losses realized: Number of realized losing positions divided by realized losing positions plus currently held positions with losses.
Without the disposition effect, the measures should be closer. In Odean's analysis, PGR was around 1.5 times PLR: gains were realized about 1.5 times as often as losses. If your own PGR is clearly above PLR, the record points to exit rules worth changing.
Move exit decisions into rules to remove the asymmetry
The effect does not operate before entry, when profit and loss is still zero and neither gain-side risk aversion nor loss-side risk seeking has a reference point. Move exit decisions to that moment. Set stop and target before profit and loss begins moving, leaving less room for asymmetric emotion later.
- Place exits at entry: Submit a stop at 1R and a first target as an exchange OCO order when opening. Do not decide the exits afterward.
- Restrict stop movement: Move the stop only toward break-even or profit, never farther into loss.
- Adjust gains through trailing rules: If you want to lock in gains, use a trailing stop based on the average range of the last 20 bars (ATR) rather than discretionary early exits.
- Record outcomes: Note holding time and realized R at every exit.
- Review weekly: Compare mean holding time for winners and losers and calculate PGR/PLR; revisit exits if the gap widens.
The aim is to separate exits from emotion so live trading retains the outcome distribution the backtest assumed.
Two pitfalls
“I will hold only until break-even.” Keeping a loser until it returns to entry simply disguises the loss-side disposition effect as a rule. Entry price is personal; the market does not know it. Base stop placement on structure and volatility.
“Taking partial profits solves everything.” Scaling out can reduce volatility and emotional pressure while profitable, but in a trend system it may cut the right tail of large winners and lower average gains. Introduced to reduce the disposition effect, it can sometimes reproduce the same expectancy cost. Backtest separately where partial profit-taking helps and where it hurts.
Exit habits become visible only in the record
The disposition effect is hard to see in entry analysis. Win rate can look healthy, signals can be correct and no single dramatic mistake appears, yet the account slowly declines. The missing information is in exits, which become measurable in a trade table. Record holding time and realized R for every fill; regularly compare winners and losers and PGR with PLR. “I close winners too early” then becomes a measure you can change. The first step is not stronger resolve. It is measuring your own exits.