You were stopped out by a handful of pips. The trade then turned around and covered the entire distance you had originally marked as the target, without you in it, and your risk was managed perfectly the whole way down.
That is not bad luck happening twice a month. That is a stop sitting in the wrong place, and there is data on your own hard drive that says exactly where the right place is.
By the end of this you will be able to pull one number out of every trade you have already closed and use it to decide whether your stops are protecting the account or quietly funding the other side of the market.
Key Findings
- Definition: maximum adverse excursion is the furthest a trade travelled against you between entry and exit, recorded whether the trade finished as a win or a loss.
- Why it matters: profitable trades tend to dip only a shallow distance before turning, so a stop placed inside that band systematically closes trades that were correct.
- Origin: John Sweeney set the method out in his 1996 Wiley book Maximum Adverse Excursion, where he argued that stop distance belongs to measurement rather than preference.
- Prerequisite: the analysis is only valid if the entry price in your journal is a price you could genuinely have been filled at.
What is maximum adverse excursion?
The deepest unrealised loss a trade carried before it closed. Measure from your entry to the worst price the market printed against you while you held it, and record that distance. The outcome is irrelevant to the measurement. A trade that finished up two hundred points after first sinking forty against you has an adverse excursion of forty.
Sweeney’s argument in 1996 was blunt and has aged well: traders choose stop distances the way they choose a lucky shirt, then never audit the decision. Yet every closed trade has already run the experiment. The record is sitting there.
Log that number for a run of trades from the same setup and something separates. Winners cluster shallow. They go a little against you, then work. Losers keep travelling, because whatever the entry was reading turned out to be wrong and nothing stopped price continuing.
Why is a round-number stop usually the wrong one?
Because it was chosen before the trade and never checked afterwards. Twenty pips because twenty is tidy. Fifty because the account can afford it. Neither number knows anything about how this setup on this instrument actually behaves in the first hour after entry.
An ATR-based stop is a real improvement on that, and I use one as a floor. It at least scales with current volatility instead of a habit. But average true range describes the noise of the market. Adverse excursion describes the noise of your entry inside that market, which is a narrower and more useful thing.
| Stop method | What it is based on | Adapts to volatility | Uses your own results | Main weakness |
|---|---|---|---|---|
| Fixed pip or round number | Habit, or what the account can absorb | No | No | Unrelated to how the setup behaves |
| Percentage of account | Position sizing convenience | No | No | Puts the stop where the maths is neat, not where price is |
| ATR multiple | Recent range of the instrument | Yes | No | Measures market noise, not entry noise |
| Structure or swing low | The chart’s own levels | Partly | No | Reliable, but the level can sit much further than needed |
| Adverse excursion band | The recorded behaviour of your winners | Yes, as the sample updates | Yes | Needs a real sample and an honest journal |
How do you build the table from your own trades?
Add two columns to whatever you already keep. Most trading journals record entry, exit and result, which tells you nothing about the middle of the trade.
Record the worst price against you and the best price in your favour for every position. Then filter to winners only, sort by adverse excursion, and read down the column until the numbers stop being crowded together and start being outliers. That gap is the answer you came for.
What does the cluster tell you to change?
Put the stop just beyond the depth that most of your winners survived, and no further. If your current stop sits inside that band, you are paying a fee for being right. If it sits well outside it, you are carrying risk that none of your winning trades ever needed.
That second case is the one traders miss, and it costs more than it looks. Risk per trade is stop distance multiplied by position size. Trim a stop that was never earning its width and the same cash risk buys you a larger position on the identical setup. Same account exposure, more of the move captured.
Be honest about the trade-off in the other direction. A wider stop is not free either: it buys survival with a lower reward-to-risk ratio on every trade, and it will feel worse to sit through. The point is not to widen or tighten by reflex. It is to stop guessing and let the position sizing follow a distance you can defend.
RelicusRoad Pro
Have you been trading for a while but have never made consistent profits or are you new to FOREX trading and want to get a head start? Try RelicusRoad and you'll never look back.
Get RelicusRoad ProMAE or MFE: which one are you actually asking about?
Two edges of the same trade, answering two different questions.
| Measure | What it records | What it should set |
|---|---|---|
| Maximum adverse excursion | Deepest move against you before the exit | Stop distance, and when a trade has left the population of your winners |
| Maximum favourable excursion | Best unrealised gain before the exit | Target distance, and whether you are exiting too early or asking for too much |
If your winners routinely offer far more than you take, the problem is not the entry and no new indicator will fix it.
What quietly invalidates the whole exercise?
Three things, in ascending order of damage.
A sample drawn from a single stretch of market conditions describes that stretch, not your strategy. Change the rules partway through a sample and you have blended two different populations into one average that represents neither.
The serious one sits at the entry. Every excursion is measured from an entry price, so if that price was never available to you, everything downstream is arithmetic on a number you invented. This is how a signal that shifts position after the candle finishes ruins more than a backtest : the arrow settles somewhere flattering, your journal records the flattering price, and the adverse excursion you calculate is smaller than the one you actually lived through. The same mechanism that makes repainted backtests look profitable also makes your stop research recommend a stop too tight to survive.
What does this ask of the tool on your chart?
One thing, and it is not accuracy. It is that the entry it gave you yesterday is the entry it still shows today.
An arrow that locks at candle close gives you a fixed reference point, which is the entire foundation this analysis rests on. RelicusRoad Pro is built to that rule across MetaTrader and TradingView, so the entries you log are entries you could have been filled at, and the excursion column you build from them means something a month later.
It will not tighten a drawdown by itself. Nothing on a chart decides your size, your patience, or whether you honour the stop once it is placed. What a fixed reference point does is make your own trade record admissible as evidence, and once you manage risk from real numbers rather than round ones, stop placement stops being a personality trait.
Frequently asked questions
What is maximum adverse excursion in trading? It is the worst point a trade reached against you between entry and exit, measured as a distance from your entry price. A trade that closed for a gain after first dropping thirty points against you has an adverse excursion of thirty. The final result does not change the measurement, and the value exists for every trade you have ever closed.
How is MAE different from MFE? They measure opposite edges of the same trade. Adverse excursion is the deepest move against you; maximum favourable excursion is the best unrealised profit the trade offered before you closed it. The first informs where the stop belongs. The second informs whether your profit targets are leaving money on the table or asking for a move the setup rarely delivers.
How many trades do you need before MAE data is worth acting on? Enough winning trades from one consistent setup that a cluster is visible rather than implied, and enough calendar time that they were not all taken in the same conditions. There is no magic count. The practical test is stability: add the next batch and see whether the shape of the distribution moves. If it barely shifts, you have something.
Does maximum adverse excursion work for scalping as well as swing trading? Yes, though the sample builds faster on short timeframes and decays faster too. Spread and slippage make up a much larger share of a small excursion, so measure from your real fill price rather than the signal price, or the shallow end of your distribution will flatter you.
Can MAE analysis tell me when to exit a losing trade early? It can flag the point at which a trade has stopped behaving like your winners did, which is not a prediction. If almost none of your profitable trades went beyond a certain distance against you, a position well past it is no longer in the population you studied. Cutting there is defensible, but you are trading a smaller average loss for a lower win rate, so test it on the record first.
Add the two excursion columns to your journal this week, and if you want entry prices that stay put long enough to be worth measuring, start with RelicusRoad Pro .