Socius NewsYou closed the winner at 0.6R because it felt like enough. You kept the loser at -1.4R because it might come back. Both decisions felt reasonable in the moment. Together they have a name, a measurement and four decades of literature behind them.
The disposition effect in trading is the tendency to sell winning positions too soon and to hold losing positions too long, so that realized gains stay small and frequent while realized losses stay rare and large. Shefrin and Statman named it in 1985. Odean measured it in 1998 on 10,000 brokerage accounts: a paper gain was about 50 percent more likely to be sold than a paper loss. Your own trade history can measure it too.
What is the disposition effect in trading?
Shefrin and Statman published "The disposition to sell winners too early and ride losers too long" in the Journal of Finance in 1985. They built the idea from four parts: prospect theory, mental accounting, regret aversion and self-control. You open one mental account per trade. The account is priced against your entry. A gain is a closed chapter you want to bank. A loss is an open chapter you refuse to end.
The effect has two halves and you need both to name it. The winner half: you exit at a small positive R, often well before your planned target. The loser half: you let the trade run past the level where you said you would leave. Neither half alone is the disposition effect. A trader who takes small wins and small losses is scalping. A trader who takes small wins and large losses is disposed.
If your median winner lasts 18 minutes and your median loser lasts 47 minutes, the asymmetry is in your data before it is in your psychology.
14.8%
Proportion of paper gains realized (PGR)
10,000 accounts, 1987 to 1993
9.8%
Proportion of paper losses realized (PLR)
same accounts, same days
1.5x
A gain was 50 percent more likely to be sold than a loss
December excepted
What does Odean's 10,000-account study actually show?
Terrance Odean took the trading records of 10,000 accounts at a large discount broker from January 1987 to December 1993. On every day an account sold something, he counted every position that could have been sold at a gain and every position that could have been sold at a loss. Then he counted which ones were actually sold.
The proportion of gains realized, PGR, came out at 0.148 for the full year. The proportion of losses realized, PLR, came out at 0.098. The ratio is about 1.5. A stock sitting in profit was more than 50 percent more likely to leave the account than a stock sitting at a loss. In December the pattern reversed: PGR fell to 0.108 and PLR rose to 0.128 as investors sold losers for tax reasons. The one month with an external reason to sell losers was the one month people did it.
The second finding matters more for you. The winners that were sold earned an average excess return of 2.35 percent over the following year. The losers that were kept returned -1.06 percent. The gap is about 3.4 percent. On average the position people chose to close was the one worth keeping, and the position they chose to keep was the one worth closing.
Barber, Lee, Liu and Odean repeated the test on the Taiwan Stock Exchange from 1995 to 1999: nearly four million traders and over one billion trades. Eighty-four percent of investors sold winners at a faster rate than losers. Individuals showed a PGR of 9.43 percent against a PLR of 2.33 percent. Mutual funds and foreign institutions did not show the effect at all. The bias is not a property of markets. It is a property of the person holding the position.
How much more readily each group sold gains than losses
PGR divided by PLR, Taiwan Stock Exchange, 1995 to 1999
Source: Barber, Lee, Liu and Odean (2007), European Financial Management
Why do you sell winners too early?
Kahneman and Tversky published prospect theory in Econometrica in 1979. Three of its features explain the whole pattern. You evaluate outcomes against a reference point, not against your total wealth. The value function is concave for gains and convex for losses. And losses loom larger than gains. Tversky and Kahneman's 1992 estimate put the loss aversion coefficient at about 2.25: a loss of one unit is felt like a gain of 2.25 units.
Now put yourself in each half of the trade. In a gain you are on the concave side. Each extra tick adds less satisfaction than the last one, and the risk of giving it back feels expensive. You become risk averse and you close. In a loss you are on the convex side. Each extra tick of loss hurts less than the previous one, and the chance of returning to break-even is worth a gamble. You become risk seeking and you hold. The reference point that drives both decisions is your entry price, a number the market does not know.
Shefrin and Statman added regret. Closing a loser converts a paper loss into an admitted error. Holding it keeps the error hypothetical. The trade stays open so that the mistake can stay open too.
The same convexity shows up inside a single session. Coval and Shumway studied 426 Chicago Board of Trade locals over 1998. Traders with morning losses had a 31.2 percent probability of taking above-average risk in the afternoon, against 27 percent for traders with morning gains. They placed more trades, larger trades and carried more inventory. That is the loser half of the disposition effect compressed into one day, and it is the same mechanism you will find in revenge trading.
Does the disposition effect cost money?
Odean's 3.4 percent gap is a stock market number on a one-year horizon. For a short-term trader the cost shows up somewhere else: in the shape of your R-multiples. Cutting winners caps your average win. Riding losers stretches your average loss. Those two numbers, with your win rate, are your trading expectancy.
Take an illustrative arithmetic case, not a forecast. A trader wins 55 percent of the time. Disposed exits give an average winner of 0.6R and an average loser of -1.4R. Expectancy is 0.55 times 0.6 minus 0.45 times 1.4, which is -0.30R per trade. Now keep the same entries and the same 55 percent, but let winners reach 1.2R and stop losers at -1.0R. Expectancy becomes 0.55 times 1.2 minus 0.45 times 1.0, which is 0.21R per trade. Nothing changed about the setups. The only thing that changed was which half of the trade you were willing to sit through.
There is a second cost. Barber and Odean followed 66,465 households from 1991 to 1996. The average household turned over more than 75 percent of its portfolio each year. The most active fifth earned 11.4 percent a year net against a market return of 17.9 percent. Cutting winners early is a turnover machine: every closed winner is a new decision to make, a new spread to pay and a new chance to re-enter worse. The loser, meanwhile, sits still and costs nothing until it costs everything.
“The position you chose to close was, on average, the one worth keeping. The position you chose to keep was the one worth closing.”
What does a disposition effect example look like in a trade log?
You need four columns: entry, exit, planned exit and holding time. Here is what one disposed morning looks like, on illustrative demo data.
09:31
Long EURUSD, 1R risk defined
Stop 12 pips below entry, target 2R at 24 pips. Plan written.
09:52
Closed manually at 0.6R
Price paused at 7 pips. Reason logged: felt like enough. Target was 2R.
10:40
Long again, same 1R
Same stop distance. Same target.
11:15
Stop moved 8 pips lower at -0.9R
Reason logged: it will come back. New risk is 1.7R.
13:05
Stopped at -1.9R
Session result: -1.3R on two trades with a 50 percent win rate.
Four signatures repeat across disposed accounts. First, losers are held longer than winners, measured by median holding time. Second, realized winners cluster just past break-even, between 0.2R and 0.8R, while the planned target sits at 1.5R or 2R. Third, the loss distribution has a thin far tail at -1.5R to -3R that a fixed stop would not have produced. Fourth, exit reasons split cleanly: winners are closed manually, losers are closed by the stop, or by the margin call. Any one of these is noise. All four together are the disposition effect.
What your journal would show
The following figures are illustrative demo data, not a customer account. A trader connects a cTrader account with 312 closed trades over eleven weeks. The journal groups exits in R and by holding time. Median holding time is 18 minutes for winners and 47 minutes for losers. Average winner is 0.52R. Average loser is -1.38R. Sixty-one percent of winners were closed manually before 0.8R while the logged target was 2R. Nineteen percent of losers had their stop widened at least once. Expectancy over the period is -0.24R per trade on a 56 percent win rate.
18 min
Median holding time, winners
illustrative demo data
47 min
Median holding time, losers
illustrative demo data
0.52R
Average winner
target was 2R
-1.38R
Average loser
planned stop was -1R
None of these numbers requires a new indicator. They come from executed trades that your broker already stores. Socius Trades imports the history from cTrader and MetaTrader 4 or 5, computes R-multiples from your stop distance, and splits results by holding time, hour, instrument and day. You can ask Socius AI a plain question such as "how long do I hold losers compared with winners?" and see the answer drawn from your own log. The Essential plan is 129.99 euros a year with a 14-day trial, and the platform reads your accounts without ever placing, closing or modifying a trade. Plans and limits are on the Socius Trades pricing page. We look. We never touch.
What can you actually do about it?
The literature is clear on the diagnosis and quiet on the cure, so the rules below are procedural, not promises. They make the bias visible and expensive to repeat.
Write the exit before the entry, in R. A target of 2R and a stop of -1R written at 09:30 is a contract. An exit at 0.6R at 09:52 is a breach, and it gets logged as one, with the same weight as a moved stop.
Record an exit reason on every trade: target, stop, time, or manual. After 50 trades, count how many winners were manual and how many losers were stop. If manual dominates the winners and stop dominates the losers, you have Odean's ratio in your own account.
Measure one number weekly: median holding time of losers divided by median holding time of winners. Above 1.5, the loser half is active. Below 1, you are at least not riding.
A loser never gets a wider stop. Widening converts a -1R plan into a -1.7R reality and moves your reference point further from the market. If the level was wrong, the trade was wrong.
Change the reference point. Prospect theory says you anchor to entry. Anchor to the plan instead: a trade sitting at 0.6R with a 2R target is 1.4R short, not 0.6R ahead.
Once a month, do what Odean did. Take the ten winners you closed early and the ten losers you held. Look at what the price did in the following hour. You are doing it because the gap between the two lists is the price of the bias, in your own instrument, in your own R.
Trading involves risk of loss. Socius Trades is an analytics tool, not investment advice.