MAE and MFE

A closed trade tells you one thing: what it paid. It does not tell you what it did. A trade that went straight to target and a trade that spent six hours a hair from your stop before turning around are the same row in a P&L, and they are not the same trade.

MAE and MFE recover the missing part. Between entry and exit, MAE is the worst price the trade saw and MFE is the best. Expressed in R — multiples of the amount you risked — they are comparable across instruments, position sizes and account balances, which is what makes them worth storing rather than glancing at.

  entry  2312.0    stop 2304.0    risk = 8.0 points = 1R
  worst  2305.6    ->  MAE = 0.80R  (came within 1.6 points of the stop)
  best   2339.2    ->  MFE = 3.40R
  exit   2325.0    ->  result = +1.63R

  the idea reached 3.4R; the exit kept 1.63R

What each one separates

MFE separates the idea from the exit. A trade that reached 3.4R and closed at 1.63R had an idea that worked and an exit that captured less than half of it. Whether that is a problem depends on whether the pattern repeats: a single trade is an anecdote, forty of them with the same shape is a finding about your exits.

MAE separates the idea from the stop. If your winning trades routinely dip to 0.8R against you first, your stop is doing almost no work — it is close enough to the noise that a slightly worse fill would have converted winners into losers. If your losing trades never came close to your target, the stop was not the constraint at all.

How the numbers are produced

Most journals cannot show you these, because a broker statement contains fills and not the path between them. The excursion has to be reconstructed by walking the candles that elapsed between entry and exit and taking the extremes.

That reconstruction is only as good as its resolution. Walked on 1-minute candles it is close to the truth; walked on hourly candles it will miss most of what happened inside each bar. Sage reconstructs from 1-minute data where that data exists for the symbol and period, and where it does not, the figure is absent rather than estimated — a blank and a zero mean different things, and a zero here would read as "never went against you".

Using them without over-reading them

The temptation with MAE is immediate: find the excursion that most winners stayed within, and move every stop there. It is the wrong move, for a reason worth internalising. The excursions you can see belong to trades that survived the stop you actually had. Trades that were stopped out have an MAE of exactly 1R by construction, and you cannot know what they would have done afterwards.

Tightening a stop also changes the position size that keeps your risk constant, which changes every R figure downstream. The honest way to use an MAE distribution is as a reason to test a different stop rule going forward — in replay, on trades the rule generates — rather than as a retrospective optimisation of trades that have already closed. The backtesting guide covers why the difference matters so much.

MFE deserves the same caution in the other direction. A trade that reached 3R does not mean 3R was available to you: reaching a level and offering you a fill on the way back are different events, and a target further out is reached less often. The risk/reward calculator shows the win rate a longer target would have to sustain to break even, which is usually the number that settles whether the extra R was worth reaching for.

Where traders disagree

This vocabulary is not standardised. These are the live disagreements — worth knowing about before comparing your results with anyone else’s.

• ?What candle resolution is fine enough for a reconstruction to be trustworthy, and whether an estimate from coarse data is better than no figure at all.
• ?Whether excursion should be measured to the wick or to the close — wicks include prices that may never have been fillable in size.
• ?Whether MAE distributions from surviving trades can say anything about stop placement, given that stopped-out trades are censored at exactly 1R.
• ?Whether MFE should be measured to the exit or to the end of the session, for trades closed early.

Test it on your own trades

Whether this holds is a question about your record, not about the term. These are the steps that make the answer trustworthy.

1. 1Read MAE and MFE in R rather than in points or currency, so trades across symbols and sizes are comparable.
2. 2Look at the distribution rather than the average — the shape carries the information and one outlier moves a mean a long way.
3. 3Split by winners and losers before drawing anything; the two populations answer different questions.
4. 4Treat an MAE-derived stop idea as a rule to test forward in replay, never as a re-scoring of trades that already closed.
5. 5Check how many trades have a reconstructed figure at all; a distribution built from a third of your record is a finding about that third.

You will also need

• Liquidity sweepPrice trading beyond an obvious swing high or low — often a level with equal highs or lows — and then returning back through it.
• Trading sessionThe hours during which a major financial centre is open, used to divide a 24-hour market into periods that behave differently.
• Premium and discountA range divided at its 50% level — price above the midpoint is premium, below it is discount, and the midpoint itself is equilibrium.