The Disposition Effect in Your Own Data
The disposition effect is the tendency to sell winners too early and hold losers too long. It appears in professional and retail investors alike, and in forex it shows up as a distinctive pattern in the numbers.
To test whether it affects you, compute three things from your journal. First, average win in R compared with average loss in R. If your wins average 0.8R and your losses average 1.2R, you are cutting winners and letting losers run beyond the plan. Second, the share of trades where you exited before the target. Third, the share of losing trades where you moved the stop or removed it.
Some version of this appears in most traders' records. The reason is emotional: realizing a gain feels good, and realizing a loss feels painful, so people act to lock in the good feeling and avoid the bad.
Countermeasures work best when they are mechanical. Place stops and targets as orders when you enter, so exits happen without a decision. Limit changes to stops to one direction only, toward reducing risk. Predefine partial profit rules. Review weekly with the numbers in front of you.
Reducing position size can also help, because lower stakes weaken the emotional pull.
Track your average win and loss in R weekly, and watch the ratio change as discipline improves.
Practical step: calculate your average win in R and average loss in R, and write the difference as a target for improvement.
The Disposition Effect in Your Own Data
The Disposition Effect in Your Own Data
It’s Fairman 
Re: The Disposition Effect in Your Own Data
Setting stop and target as orders at entry fixed most of this for me. When the exit is already on the server, there's less to fiddle with. The part that took longer was resisting the urge to close early when price got close to target and pulled back a bit. What helped there was hiding the profit figure on the platform, or at least not watching it, and checking the trade only at set times, like every 15 minute candle close. Staring at the open profit as it changes is exactly what triggers the disposition effect. If you only look at the chart structure, it's easier to let the plan do its job.
It’s Fairman 
Re: The Disposition Effect in Your Own Data
Moving stops only toward less risk is a simple rule that works. Never further away, not even once just for this trade. The just this once trade is usually the one that becomes your biggest loss of the month. If you want to enforce it, add a flag in the journal for every stop moved against the plan and add up the extra loss those trades caused. Seeing a number like minus 6R over a quarter from moved stops alone makes the rule feel much less optional. Some platforms and trade managers let you lock a stop so it can only move one way, which takes the decision off the table completely.
It’s Fairman 
-
FXS Prop Desk
- Posts: 1689
- Joined: Wed Oct 07, 2026 9:52 am
Re: The Disposition Effect in Your Own Data
Compare outcomes when the stop was adjusted against outcomes when it was left alone. If moved-stop trades have a much worse average result, it's a clear sign of the disposition effect, and the solution can be as simple as not allowing stop changes against the position.
Setting the target and stop at entry, and not changing them, is a simple way to reduce the disposition effect.
Setting the target and stop at entry, and not changing them, is a simple way to reduce the disposition effect.
FXS Official Team | The Idea Lab & Strategy
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: The Disposition Effect in Your Own Data
FXS Prop Desk's moved stop comparison and Fairman's 0.8R against 1.2R example fit together nicely if you put both on the same running total. The win rate can stay identical and the exits alone decide the sign.
The attached chart shows a made up sequence of 40 trades with a 1R stop, a 1.5R target and exactly 20 winners. With wins closed early at an average of 0.8R and losers stretched to 1.2R, the running total ends at minus 8R, while the same entries managed as planned finish at plus 10R.
That 18R gap comes entirely from exits, which is why the mechanical fixes in this thread work better than willpower. Nothing about entry quality changed between the two lines.
The caveat is that the plus 10R line flatters the plan. Some of those early exits would have reversed and hit the stop, so a fair comparison checks where price went after each early close rather than assuming every one reached target.
Say 3 of the 20 early winners would have touched the stop first. The plan then shows 17 wins and 23 losses, which is plus 2.5R, still far ahead of minus 8R but nowhere near the clean plus 10R.
The attached chart shows a made up sequence of 40 trades with a 1R stop, a 1.5R target and exactly 20 winners. With wins closed early at an average of 0.8R and losers stretched to 1.2R, the running total ends at minus 8R, while the same entries managed as planned finish at plus 10R.
That 18R gap comes entirely from exits, which is why the mechanical fixes in this thread work better than willpower. Nothing about entry quality changed between the two lines.
The caveat is that the plus 10R line flatters the plan. Some of those early exits would have reversed and hit the stop, so a fair comparison checks where price went after each early close rather than assuming every one reached target.
Say 3 of the 20 early winners would have touched the stop first. The plan then shows 17 wins and 23 losses, which is plus 2.5R, still far ahead of minus 8R but nowhere near the clean plus 10R.
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: The Disposition Effect in Your Own Data
One version of the disposition effect hides behind a rule that looks sensible, which is the partial close. Taking half off at 0.5R and letting the rest run to a 1.5R target averages exactly 1.0R on a full winner, so a third of the target disappears before the trade has really started.
That is fine if planned and tested, but it often starts as a way to feel the win early. The check is whether the level was written down before entry or picked while watching the open profit.
The mirror image is worth a look too. Some traders, after reading about this effect, start closing losers at minus 0.6R whenever price gets near the stop, which feels disciplined but is the same emotional exit pointed the other way.
Say 10 of those early loss cuts happen in a made up month and 4 of the trades would have turned and reached target. Saving 0.4R on the other six is worth 2.4R, but swapping 4 wins of 1.5R for losses of 0.6R costs 8.4R.
So the test FXS Prop Desk suggested for moved stops applies here too. Compare trades closed by hand against trades left to the orders, in both directions.
That is fine if planned and tested, but it often starts as a way to feel the win early. The check is whether the level was written down before entry or picked while watching the open profit.
The mirror image is worth a look too. Some traders, after reading about this effect, start closing losers at minus 0.6R whenever price gets near the stop, which feels disciplined but is the same emotional exit pointed the other way.
Say 10 of those early loss cuts happen in a made up month and 4 of the trades would have turned and reached target. Saving 0.4R on the other six is worth 2.4R, but swapping 4 wins of 1.5R for losses of 0.6R costs 8.4R.
So the test FXS Prop Desk suggested for moved stops applies here too. Compare trades closed by hand against trades left to the orders, in both directions.
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: The Disposition Effect in Your Own Data
Backtesting on tick data seems like the gold standard. The quality of the result, though, depends on where the ticks came from.
Brokers record their own quotes, which reflect their liquidity mix and markup. Independent vendors collect data from banks or ECNs, giving a different view of the market. Some free datasets are built from minute bars through interpolation, so the ticks are artificial.
Each source has gaps and errors. Spikes that never traded, missing periods, and timestamp inconsistencies are common. A strategy sensitive to short-term price moves can behave very differently on different datasets.
The best practice is to test on data that resembles the execution environment, ideally from the same broker. Where that is not possible, comparing results across two sources shows how fragile the strategy is.
Cleaning the data to remove obvious errors, such as bid prices above ask or abrupt jumps with no follow-up, is also worth doing.
Which data source do you use for backtesting scalping systems, and how did you verify its quality?
Brokers record their own quotes, which reflect their liquidity mix and markup. Independent vendors collect data from banks or ECNs, giving a different view of the market. Some free datasets are built from minute bars through interpolation, so the ticks are artificial.
Each source has gaps and errors. Spikes that never traded, missing periods, and timestamp inconsistencies are common. A strategy sensitive to short-term price moves can behave very differently on different datasets.
The best practice is to test on data that resembles the execution environment, ideally from the same broker. Where that is not possible, comparing results across two sources shows how fragile the strategy is.
Cleaning the data to remove obvious errors, such as bid prices above ask or abrupt jumps with no follow-up, is also worth doing.
Which data source do you use for backtesting scalping systems, and how did you verify its quality?