MAE, for maximum adverse excursion, is the deepest unrealised loss a trade endured before closing. It is the statistic answering a question nothing else asks: how far did it pull against you before it worked?
The levels of a trade
Contents
MAE measures the depth of the dip a trade went through before ending. Like MFE, it restores part of the path a trade statement erases, but it informs a different decision, and that is what justifies its own page.
That decision is stop placement. A stop is placed somewhere you hope is far enough to let the trade breathe, and close enough not to pay too much when you are wrong. Nobody knows where that point is, and MAE is the only measure allowing you to find it from your own activity rather than from a general rule.
The reasoning fits in one sentence: look at the MAE of your winning trades. That is the dip your good trades went through before leaving. A stop placed closer than that dip would have turned those winners into losers, and that is exactly the information needed before tightening anything.
Trade MAE = |most unfavourable price reached − entry price| — Average MAE = mean across all trades
MAE is a distance, so always positive. It is sometimes written negative by display convention, which changes nothing in the calculation.
On a stopped-out trade, MAE equals by construction the distance to the stop: that trade teaches nothing, and that is why the reading is done on winners.
As for MFE, candle resolution decides precision. A thirty-second dip does not exist on hourly candles, and yet it is the one that would have hit your stop.
| Average MAE of winners | What it implies for your stop |
|---|---|
| far below the stop | Your stop is wide for nothing. Tightening it would cut risk without cutting winners. |
| about half the stop | A comfortable and classic safety margin. Nothing to change. |
| close to the stop | Your winners brush the stop. Tightening by a pip would kill some of them. |
| beyond the stop | Impossible on a stopped-out trade, so these are trades where the stop was not respected. |
| very scattered | An average no longer suffices: it is the extreme value to look at, not the centre. |
The fourth row is an alarm disguised as a statistic. An MAE larger than the distance to the stop is arithmetically impossible if the stop was respected: the trade would have been closed first. Its presence therefore says a stop was moved or removed along the way, and that is a discipline fact far more important than the stop setting itself.
The last row concerns spread, and it is underrated. An average MAE of 19 pips can hide winners that went through 8 pips and others through 29. Placing your stop at the average means accepting to cut half of what works. It is the high value, not the centre, that decides a stop placement.
This winner went to within 22 pips of a stop placed at 30. A 20-pip stop would have killed it. Across all winners, the 19-pip average MAE says that tightening the stop to 25 pips would stay workable, and that going down to 15 would cut a good share of what works. That is a risk decision taken on data rather than on a feeling.
This is the use that justifies the whole measure. A tighter stop reduces the loss per trade and increases the number of stopped-out trades: the trade-off is known, what is always missing is the second term. The MAE of winners gives it exactly, trade by trade.
The method is mechanical: count how many of your winners have an MAE larger than the stop you are considering. Those are the ones you would lose. If it is two out of forty, the deal is good; if it is fifteen, it is not.
An MAE deeper than the distance to the stop cannot exist if the stop worked. Those trades stand out immediately, and they require no interpretation: the stop was moved, or it did not exist.
It is a discipline use rather than a setting one, and it is often more profitable than the other. A single stop not held can cost more than several months of placement optimisation.
The first flaw is reading MAE across all trades. Stopped-out trades have by construction an MAE equal to the distance to the stop, so they drag the average towards that value and completely mask the information. The average that matters is that of winners, and almost no tool separates it spontaneously.
The second is taking the average for a bound. A stop placed at the winners' average MAE cuts by definition roughly half of them, since half did worse than the average. Placement is decided on the high end of the distribution, not its centre, and that confusion is expensive.
The third is data resolution, as for MFE but with a different consequence. An MAE computed on hourly candles misses the thirty-second dip which would in fact have triggered your stop. Here the understatement does not produce a regret, it produces a risk decision that is too optimistic.
The classic mistake. Placing your stop at the average MAE of your winners. Half of the winners, by definition, dipped below that average: that stop would have killed them. Placement is decided on the tail of the distribution, the point beyond which only two or three trades remain, never on its centre.
| Indicator | Formula | What it measures | When to use it |
|---|---|---|---|
| MAE | worst point reached − entry | The dip endured before the end | Deciding where to place a stop |
| MFE | best point reached − entry | The peak reached before the end | Deciding when to exit |
| MAE in R | MAE ÷ risk taken | The same dip, in risk units | Comparing trades with each other |
| Worst winner MAE | the deepest MAE of a winner | The trade that came closest | Spotting luck |
Stopped-out trades have an MAE equal to your stop, by construction. Including them crushes the average onto that value and makes all the information disappear. The question asked is 'how far did my good trades dip', and it concerns them alone.
A stop placed at the winners' average MAE cuts about half of them, since half did worse. Look instead at the MAE beyond which you find only two or three winners: that is where the stop becomes defensible.
They are arithmetically impossible if the stop held. Each one is therefore a trade where it was moved or removed, and that list almost always pays more than optimising placement. It is a discipline check disguised as a statistic.
MAE stands for maximum adverse excursion. It is the deepest unrealised loss a trade endured between opening and closing. A trade that ended as a winner after going through −22 pips has an MAE of 22 pips.
To decide where to place a stop, from your own trades rather than from a general rule. The MAE of your winning trades says how far your good trades dipped before leaving: a stop placed closer would have turned them into losers.
No, only the winners. A stopped-out trade has by construction an MAE equal to the distance of your stop, so including them crushes the average onto that value and removes all the information. The question only concerns trades that ended up working.
No, and that is the most frequent mistake. Half your winners dipped below the average, by definition: that stop would have cut them. Placement is decided on the high end of the distribution, the point beyond which only two or three trades remain.
That the stop was not respected. It is arithmetically impossible otherwise: if the stop had worked, the trade would have been closed before reaching that point. Those trades signal a stop moved or missing, and that is a discipline fact more important than any setting.
Like MFE, MAE is recomputed by Tradoshi from downloaded historical candles, because no broker statement contains a trade's path. The calculation is cached and trades with no data available are dropped rather than counted as zero, which avoids dragging the average down.
Average MAE (worst moment) is a metric in the Reports screen, crossable with any dimension. The one that serves here is Side, long and short separately, and above all Symbol: a stop that suits a currency pair has no reason to suit an index, and that crossing shows it in one line.
To go deeper on this: read the full guide. Every term is gathered in the glossary.