Every trader has heard the instruction: cut your losses, let your profits run. Almost nobody checks whether they do it. The check takes one number, and most journals do not report it because they average the two halves together and hand back a single hold time that conceals the very thing you wanted to know.
One number, two behaviours
Suppose your journal tells you the average trade is held for twenty-six minutes. That is a fact about your record and it answers nothing. It is equally consistent with a disciplined trader whose winners and losers both close in twenty-six minutes, and with one who takes profits in eight minutes and sits in losses for forty-five, hoping.
Those two traders have entirely different problems. The first may have no problem at all. The second is systematically giving back the edge they have — every winner clipped before it matures, every loser given room it did not earn — and their single average hold time will never once mention it.
Split it by outcome
Take the same journal and report two figures instead of one:
| Outcome | Trades timed | Median hold |
|---|---|---|
| Winners | 142 | 8m 10s |
| Losers | 118 | 45m 30s |
That is a trader who holds losing positions five and a half times as long as winning ones. No amount of arguing with the interpretation changes the shape of it: profits are being taken quickly and losses are being waited out. The behaviour is visible in a way it simply is not in "twenty-six minutes".
A winning position is comfortable to close — the profit is real, and taking it feels like discipline. A losing position is uncomfortable to close, because closing it converts a paper loss into a decided one. Holding costs nothing emotionally in the moment and everything over a year. The asymmetry is not a knowledge problem; every trader knows the rule. It is a behaviour, which is why it shows up in a measurement of behaviour and not in a strategy document.
Use the median, and look at both
Hold times are badly skewed. Forty scalps that close in minutes, and one position you sat in from the open until you were forced flat before the close, produce a mean that describes nothing you have ever done. The median — the middle trade — sits much closer to your normal behaviour.
On a prop account the outlier is usually a session rather than a weekend. The overwhelming majority of evaluation and funded plans are day-only — you must be flat before the daily close — so the longest trade available to you is bounded by the session itself. A handful of firms do permit overnight holds; if yours is one, your outliers can run longer and the mean will suffer for it more.
That bound makes the skew smaller than a swing trader's, and more revealing. A position carried for three hours inside a day of four-minute trades was a decision, not a strategy.
Report both anyway. A large gap between the mean and the median is itself a finding: it says an outlier is carrying the number, and outliers on the losing side are exactly where the damage lives.
What has to be excluded, and why it matters
A duration needs a timestamp at each end. Auto-logged and imported trades carry them; a trade typed in by hand usually does not.
A trade with no exit timestamp must be left out of the figure, not counted as zero. Counting it as instant would drag the average down — and because hand-entered trades skew toward the ones you were least happy to record, the error would land on the losing side and make your losses look shorter than they are. That is wrong in the most flattering possible direction, which is the worst way for a number to be wrong.
Each side also needs its own count. If eighteen of your losers have no exit clock and only two of your winners do, the two medians are not describing comparable samples, and the tool should say so rather than print a ratio.
A median over three trades is not a habit. Five timed trades on each side is a sensible floor for looking at all, and a few dozen before you would act on the gap. Below that, a single unusual trade moves the middle value and you are reading noise.
What the comparison does not tell you
It does not tell you that you are trading badly. It is a description of your record, not a verdict on your method.
If you trade with a fixed stop and a target that lets profits run, your winners should be the longer ones, and the reversed pattern is a genuine warning. But some approaches produce longer losers honestly — a mean-reversion position held for a move back that eventually arrives will spend much of its life underwater. Whether the gap is a leak depends on what you were trying to do, and only you know that.
What the number does is remove the option of not knowing.
Where to look next
If the gap is there, the next question is where it comes from. Grouping your trades by duration — under a minute, one to five, five to fifteen, and so on — and reading the P&L of each band answers a different question: not "how long do I hold" but "which holding periods actually make me money". The two cuts together usually locate the behaviour precisely: a band that loses steadily, full of trades held far past the point the idea stopped working.
Or have it worked out for you
Choptick reports hold time split by outcome — median winner against median loser, with the number of timed trades behind each — on the dashboard and in the export you can hand to an AI assistant. Trades with no exit timestamp are excluded and counted separately, never treated as instant.
See ChoptickFigures in the examples are invented and illustrative. Nothing here is financial advice.