Part 3: Size, signals and outcome · Chapter 10 of 14 · 8 min
The weight of the outcome
Executing the plan perfectly does not avoid losses: 54% of trades lose, 36% of the weeks with closed trades are negative (14 out of 39), and 36% of the windows of 4 consecutive trades. The result of a trade, of a week, even of a month (3 out of 10 are negative) says little about the quality of the decision. About 2,950 trades are needed to see an edge of 0.1 R with this dispersion: to judge behaviour you look at adherence, not at the result.
Outcome bias
Outcome bias is judging the quality of a decision by its outcome. A winning trade seems "well taken", a losing one "badly taken". The reasoning is understandable and wrong: in a plan that loses 54% of the time, a perfect decision (executing the plan) has more chances of losing than of winning. The result of a single trade is dominated by chance.
Two consequences, both harmful:
- After a loss from an executed plan, you think "I got it wrong" and change something (chapters 4, 6, 7, 8).
- After a win from a deviation ("I closed early and did well"; "I widened the stop and it worked"), you reinforce a behaviour that costs (chapters 6 and 7: the 57% of wins with the widened stop).
How often a well-executed plan loses
Numbers on the plan (79 trades, +0.74 ± 0.43 R, all executed as per plan):
| Unit | How many | Negative | Minimum sum | Maximum sum |
|---|---|---|---|---|
| Single trade | 79 | 54% | −1.02 R | +2.99 R |
| 4 consecutive trades | 76 windows | 36% | −4.1 R | +10.2 R |
| 10 consecutive trades | 70 windows | 7% | −2.1 R | +17.9 R |
| 20 consecutive trades | 60 windows | 0% | +4.7 R | +26.5 R |
| Week (with closed trades) | 39 | 36% (14) | −2.0 R | +9.0 R |
| Month | 10 | 30% (3) | −2.1 R | +15.6 R |
One week in three (14 out of 39) of a perfectly executed plan closes at a loss, and one window of four trades in three. In the 39 weeks with closed trades the average weekly sum is +1.49 R with a deviation of 2.91 R: the sign of a week says almost nothing. At four weeks (about a month) things improve, but three months out of ten are still near zero or negative (−0.1, −0.4, −2.1 R). Only from 20 trades on, in this series, is the sum always positive; but it is a single series with an unproven edge.
Read the chart in words
Daily gold chart with, below, bars of the weekly R of the plan: one week in three below zero, some very high.
The window of judgment
This has a practical consequence: the window in which you judge. After a week, four trades, a month, you do not have the information to say whether the plan works or whether it was executed well. To see an edge of 0.1 R per trade with a deviation of 1.94 R you need about 2,950 trades (power 80%, book 11, chapter 12); to see one of 0.5 R you need 118. And a behaviour that costs 0.1 R per trade (for example a target cut a little more often) is in the same order of magnitude: it does not show in the results for years.
So behaviour is not judged by the result. It is judged by something you can count at any moment: adherence. For each trade, was the plan executed (yes/no)? If yes, the decision was right, whatever the outcome; if no, there is a deviation, whose cost you know from the measurements of this book. The journal (chapter 2) makes this distinction by itself: one field for the outcome, one for adherence.
The four-box matrix:
| Good outcome | Bad outcome | |
|---|---|---|
| Plan executed | Merit of the plan (and a bit of luck) | Normal: it happens 54% of the time |
| Plan not executed | Luck: reinforces a behaviour that costs | The cost of the deviation, plus chance |
The two boxes that count are the bottom left and the top right: a normal loss must not be punished, a win from a deviation must not be rewarded.
After three losses in a row
In the plan, after three losses in a row, the winning share of the following trades is 6 out of 11 (55%), against 46% in general. With 11 cases you can only say "there is no signal". It is the kind of moment in which the weight of the outcome is felt; and it is the moment in which the number to look at is adherence, not the last loss.
What to do with it
- Record the adherence of every trade, before the outcome if possible.
- Evaluate the week by adherence, not by result: "did I execute all the planned trades?".
- Fix a window of judgment for the result (for example 100 trades or 6 months) and respect it.
- Do not reward a win from a deviation, and note it as a deviation even if it went well.
- Do not punish a loss from an executed plan: it is one of those (54 trades out of 100) that are part of the plan.
Procedure in four steps
- For each trade mark adherence (yes/no) and, if no, the deviation.
- Every week look at adherence (how many trades executed out of how many planned) before the result.
- Judge the result only on the window decided in advance.
- Do not change the plan in the middle of a window.
Exercises
Chapter exercises
1. A perfectly executed plan has 14 weeks out of 39 at a loss (36%). What does it mean for whoever judges from one week?
2. In 4 consecutive trades the sum is negative in 36% of cases, in 20 in 0%. What does it imply?
3. You closed early a trade outside the plan and it won. What do you do?
4. Why is behaviour judged by adherence and not by the result?
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.