Part 1: Measuring behaviour · Chapter 3 of 14 · 8 min
How much a winner hurts
A winning trade almost always passes through a moment when it looks like a loser. In the plan, half of the wins fall by at least 0.41 R before reaching the target, 36% by at least 0.5 R and 11% by at least 0.8 R (almost at the stop). Conversely, 37% of the losses (16 out of 43) had been at +1 R before being stopped. Wins last 13 hours on median, losses 6: the plan asks you to stay in while it hurts.
MAE and MFE
For every trade, the journal (chapter 2) can record two numbers, in R:
- MAE (maximum adverse excursion): how far the price went against you at most, from entry to exit.
- MFE (maximum favourable excursion): how far it went in your favour at most.
They say how the road went, not just where it ended. A win with MAE 0.8 R and one with MAE 0.1 R are both worth +2.99 R, but they are two different experiences.
How much a win suffers
In the 36 wins of the plan, the maximum adverse excursion before reaching the target:
| Measure | Value |
|---|---|
| Median MAE | 0.41 R |
| 75th percentile | 0.58 R |
| 90th percentile | 0.85 R |
| Wins with MAE ≥ 0.5 R | 36% |
| Wins with MAE ≥ 0.8 R | 11% |
| Wins with MAE ≥ 0.95 R (almost stopped) | 3% |
More than one win in three (36%) went through a moment when it was losing half a risk, and one in nine (11%) through one where it was a hair from the stop (0.8 R or more). Across all 96 rules with N = 80 the picture is similar: 16% of wins had an MAE of at least 0.8 R.
Read the chart in words
Hourly gold chart of 18-19 March 2026 with the entry, the near stop, the target and the 80-candle low of a winning sell.
| When | What | Price |
|---|---|---|
| 18 Mar 2026 20:00 UTC | entry | 4,825.31 |
| 18 Mar 2026 20:00 UTC | stop: −1 R (48.3 $) | 4,873.62 |
| 18 Mar 2026 20:00 UTC | target: +3 R | 4,680.37 |
| 18 Mar 2026 12:00 UTC | 80-candle low | 4,837.49 |
Times in UTC.
In this win (14 hours in the trade) the price rose against the position up to 0.89 R, that is about 5 $ from the stop, and then fell all the way to the target. Whoever had tightened the stop "for safety" during that phase would have been stopped out.
How much gain a loss saw
Conversely, the 43 losses of the plan, with the maximum favourable excursion before being stopped:
| Measure | Value |
|---|---|
| Median MFE | 0.61 R |
| 75th percentile | 1.24 R |
| Losses that had been at +0.5 R or more | 53% (23 out of 43) |
| Losses that had been at +1 R or more | 37% (16 out of 43) |
| Losses that had been at +2 R or more | 7% (3 out of 43) |
Almost two losses out of five (16 out of 43) had had, at some point, an unrealised gain of at least 1 R. Those 16 losses are worth −15.2 R in all. Overall, 66% of all trades reach at least +1 R, 49% at least +2 R, and only 41% reach the 3 R target.
Read the chart in words
Hourly gold chart of 23-24 February 2026 with the entry, the stop, the far target and the 80-candle high of a losing buy.
| When | What | Price |
|---|---|---|
| 23 Feb 2026 14:00 UTC | entry | 5,193.99 |
| 23 Feb 2026 14:00 UTC | stop: −1 R (33.0 $) | 5,160.96 |
| 23 Feb 2026 14:00 UTC | target: +3 R | 5,293.09 |
| 23 Feb 2026 06:00 UTC | 80-candle high | 5,178.76 |
Times in UTC.
At +1.80 R there were still 1.2 R to the 3 R target: it is the situation in which the voice "close now and take the gain" is loudest. Whoever closed there would have collected +1.8 R instead of −1.01, and would have given up the 3 R in all the other trades in which the price, from there, reached it.
Duration: the part that weighs
| Wins | Losses | |
|---|---|---|
| Median duration | 13 hours | 6 hours |
| 75th percentile | 28 hours | 15 hours |
Losses arrive early (median 6 hours, the time to touch the stop) and wins take longer (13 hours, and one in four more than 28). Whoever holds a loss understands quickly how it goes; whoever holds a win on its way must stay in while the price moves against them, for a whole day, without knowing whether it will be a win.
What follows
The two numbers together explain the temptations we will measure:
- Cutting the gains (chapter 6): 66% of trades see at least +1 R, and closing there gives a gain that seems safe.
- Tightening or moving the stop to avoid suffering: 36% of wins lose half a risk before winning (chapter 6).
- Widening the stop (chapter 7): the moment when the price is 0.8 R against is the one in which the idea of giving it room is strongest.
- Skipping after losses (chapter 4): a loss arrives early and hurts at once.
None of these choices is proof of weakness: they are understandable answers to real data. The point of the plan is to decide them beforehand, with the numbers.
What to do with it
- Record MAE and MFE of every trade, in R.
- Look at the distribution, not the single case: the win that suffered is normal, and so is the loss that had gained.
- Write beforehand what you do at +1 R and at 0.5 R against (move the stop? close? nothing?) and measure the cost of the rule (chapters 6 and 7).
- Remember the duration: if wins last a day, a plan that asks you not to touch anything for a day costs attention.
- Do not judge the trade by the moment: judge it by the plan.
Procedure in four steps
- For every trade, record MAE and MFE in R.
- Compute, for wins and losses, the distribution of MAE and MFE and the duration.
- Write the rules for the two hard moments (+1 R in profit, 0.8 R against) before you face them.
- Measure on your plan the cost of each rule (chapters 6 and 7).
Exercises
Chapter exercises
1. In the plan, what share of wins falls by at least half a risk (0.5 R) before reaching the target?
2. 37% of the losses of the plan (16 out of 43) had been at +1 R or more before being stopped. What does it say?
3. Wins last 13 hours on median and losses 6. What does it mean for whoever executes the plan?
4. A win fell to 0.89 R against before reaching the target. What would have happened with the stop tightened "for safety"?
Have a question or want to share your exercise?
Post in the community, or join the free signals room on Telegram.
Educational content, not financial advice. Trading involves risk.