The Distributions tab on the Performance page has a panel that most people scroll past because it looks like a chart rather than a number. It is the most useful thing on the page. It takes every closed trade in your current scope and plots them as a histogram, and unlike a win rate or an average it shows you the whole shape rather than one squeezed summary of it.
On the capture I am working from the panel header reads n equals 87, mean equals -0.53, median equals 0.00. Three numbers, and the third one is the interesting one.
What the panel is actually plotting
The panel has two controls. A metric selector offering PnL percent, R-multiple, hold time, maximum favourable excursion and maximum adverse excursion, and a bin count offering 10, 20, 30 and 50. It plots one metric at a time across your closed trades, with the count of trades on the vertical axis.
On the PnL percent reading here, the horizontal axis runs from about -17.34 at the far left to about 1.00 at the far right, split into twenty bins, and the vertical axis tops out at 66. So the tallest single bar holds 66 of the 87 trades, and it sits in the bin next to zero. Everything else is spread thinly to the left, and there is almost nothing to the right of the centre at all.
That is a specific and readable shape. Most trades did nothing measurable. A small number of trades did serious damage. Nothing did much good.
A median of zero and a mean of -0.53 in the same line
When the median and the mean disagree, the gap between them is the tail talking.
A median of 0.00 says the middle trade in the sorted list finished at nothing. That is the typical outcome and it is neutral. A mean of -0.53 says the average outcome was a loss. The only way both are true is if the losses are concentrated in a minority of trades that are individually much larger than the typical trade, which is exactly what an axis reaching -17.34 with a tall bar next to zero looks like.
This is a genuinely useful diagnosis because it rules things out. It is not a win rate problem in the usual sense, since your typical trade is not bleeding. It is not a costs problem, because costs would shift the whole distribution slightly left rather than produce a long tail. It is a small number of trades that were allowed to become much larger losses than the rest, and those trades are individually identifiable in the export.
Go and find them. Five or six rows will account for most of the damage, and reading those five or six with the dates and the sizes attached will tell you more in ten minutes than another month of trading will.
Three shapes and what each says about your exits
Across the accounts I have looked at, this panel produces one of three broad shapes on the PnL percent metric. These are readings of a mechanism rather than predictions, and each one implies a different fix.
- A tall centre with a thin long left tail. This is the shape on screen here. It says the exits are asymmetric in the wrong direction, letting a few positions run well past the point where the loss stopped being the one you sized for. Usually this is a stop that is not enforced, or a stop placed further away than the loss you would actually accept. It is the most mechanically fixable of the three, because the fix is a rule rather than a judgement.
- A fat right tail with many small losses and a low win rate. The trend-following shape, and it is intentional in a lot of good systems. It says exits are letting winners run and cutting losers quickly. The danger with this shape is different. You need the right tail to keep appearing, so a stretch without one is not automatically evidence the method stopped working, and cutting the strategy during that stretch is the classic error.
- A roughly symmetric spread with the mass away from zero on both sides. This says your exits are symmetric, typically a fixed target and a fixed stop at comparable distance. Nothing wrong with it, but it means the entire result depends on hit rate, so your edge has to live in the entry and you should be measuring it there.
The shape you have determines which question is worth your next hour. Shape one means go and write an exit rule. Shape two means go and check whether the right tail is still turning up at its historical rate. Shape three means go and test the entry, because nothing else in the system is generating an edge.
The bin control changes the picture, so fix it and leave it
The bin selector is a rendering choice, and on a sample this size it changes the apparent shape substantially. Eighty-seven trades across twenty bins is roughly four trades per bin on average, which means most bins are noise and one bin is carrying the picture.
Drop to ten bins and the tail smooths into something that looks less severe. Push to fifty and the tail fragments into scattered single-trade bars that look like a wider spread of outcomes than you actually had. Neither is wrong, and neither should be chosen after you have seen what each looks like.
Two practical rules. Pick one bin count and use it every time so your readings are comparable month to month. And read the extremes by counting trades rather than by looking at the silhouette. The sentence that matters is that six trades finished worse than minus ten percent, and that sentence is true regardless of how the bars are drawn.
The two metrics that are empty and why that matters
The metric selector on this panel offers five options and on this real-money account two of them have no data. The hold time histogram reads n equals zero. The excursion histograms read n equals zero with a note that maximum favourable and adverse excursion are only recorded on paper trades.
That is worth knowing before you plan an analysis around them, because those two metrics answer the questions you most want answered about exits. Maximum favourable excursion tells you how far a trade went in your favour before you closed it, which is the direct measurement of whether you are exiting winners too early. Maximum adverse excursion tells you how far it went against you before it recovered, which is the direct measurement of whether your stops are too tight or too loose.
Without them you are inferring exit quality from the shape of the outcome distribution, which works but is indirect. If exit quality is the thing you are actually trying to fix, the workaround is to run the same strategy in the paper scope alongside the live one for a period, and read the excursion histograms there. The scope buttons at the top of the page are where you switch, and the page notes in its own subtitle that the default view reflects real-money trades with paper kept separate.
The other metric to read straight after PnL percent is R-multiple, which expresses each trade against the risk you took on it rather than against notional. Comparing those two shapes is the fastest way to see whether an ugly PnL distribution is a signal problem or a sizing problem, because a distribution that looks acceptable in R and terrible in percent is telling you the trades were fine and the sizing was not.