Part 1: Measuring behaviour · Chapter 2 of 14 · 8 min
The trading journal
The trading journal is the memory that does not deceive: every trade in R, with the plan and the deviations. On our plan, 79 rows already give everything you need: mean R by direction (buys +0.79 ± 0.58 over 42, sells +0.68 ± 0.64 over 37: indistinguishable), by hour band (from +0.18 to +1.66, with 6-38 trades per band) and by month (from −2.1 to +15.6 R). But every cut reduces the trades: ten bands with 16 trades are not a discovery, they are multiple trials.
Why a journal
A trader's memory is selective: it remembers the win that could have been bigger, the loss that "should not have been taken", the skipped trades that went well. It remembers the others badly. A trading journal is a memory written at the moment, with numbers, which cannot be corrected afterwards. It serves three things: computing the measures of the plan (mean R, error, losing streaks, drawdown), discovering the deviations from the plan and their cost, and deciding with data and not with memory.
The fields
A minimal journal, for each trade:
| Field | Why it is needed |
|---|---|
| Entry date and time (UTC) | For cuts by hour and by month |
| Instrument and direction | For cuts by direction |
| Rule or setup | To compare different plans |
| Entry, stop, target | To recompute the risk and check that they are those of the plan |
| Risk in $ and in % of the account, ounces | To check the size (book 11) |
| Exit: price, time, reason (target, stop, time, manual) | To see how many exits were manual |
| Result in R, cost | To add up and compare |
| Maximum adverse and favourable excursion (MAE, MFE) in R | To see how the trade goes (chapter 3) |
| Adherence to the plan (yes/no) and which deviation | The most important field (chapter 13) |
| Note written before entry | To compare the expectation with what happened |
A journal filled in
The first eight trades of our plan, as they would stand in the journal (risk in $ per ounce; MAE only for wins, because in a loss the adverse excursion is the stop itself):
| No. | Entry (UTC) | Direction | Risk | Hours | Outcome | R | MAE | MFE |
|---|---|---|---|---|---|---|---|---|
| 1 | 22 Dec 2025, 01:00 | buy | 18.4 $ | 15 | target | +2.98 | 0.38 R | 3.37 R |
| 2 | 22 Dec, 21:00 | buy | 21.8 $ | 24 | target | +2.99 | 0.69 R | 3.31 R |
| 3 | 26 Dec, 16:00 | buy | 20.9 $ | 7 | stop | −1.01 | stop | 0.92 R |
| 4 | 29 Dec, 13:00 | sell | 37.2 $ | 38 | target | +2.99 | 0.00 R | 4.12 R |
| 5 | 5 Jan 2026, 06:00 | buy | 26.8 $ | 48 | time | +1.70 | 0.91 R | 2.95 R |
| 6 | 9 Jan, 15:00 | buy | 19.1 $ | 2 | stop | −1.02 | stop | 0.13 R |
| 7 | 11 Jan, 23:00 | buy | 25.1 $ | 14 | target | +2.99 | 0.16 R | 3.23 R |
| 8 | 12 Jan, 15:00 | buy | 24.6 $ | 5 | stop | −1.01 | stop | 0.37 R |
From these rows alone you can see what the journal allows: the third trade was, at one point, at +0.92 R before being stopped; the fifth made its peak at +2.95 R and closed at time at +1.70; the fourth went straight the right way (MAE zero). They are the data of chapters 3, 6 and 7.
What it allows to compute
With all 79 rows, the journal yields measures that no memory gives:
- By direction. Buys: 42 trades, mean R +0.79 ± 0.58; sells: 37, +0.68 ± 0.64. They cannot be told apart: each sits inside the other's error.
- By entry hour band (UTC). 0-7: 16 trades, +1.66; 7-13: 19, +0.18; 13-21: 38, +0.58; 21-24: 6, +0.99.
- By exit month. December +7.9 R, January +15.6, February +2.0, March +8.3, April −0.1, May +4.9, June +7.8, July −0.4, August +14.1, September −2.1.
Read the chart in words
Daily gold chart with, below, ten bars of monthly R of the plan, mostly positive.
Read the chart in words
Daily gold chart with, below, two cumulative R curves, one for buys and one for sells, both rising.
What the journal does not allow
The journal is a powerful tool and easy to misuse. Every cut (direction, hour, month, day of week, volatility) divides the trades into smaller groups, and each group has a larger error. The 0-7 band with +1.66 R over 16 trades has an error of about ±1: it is not "my good hour". With four bands, two directions and ten months you already have twenty boxes: some will stand high by chance (book 12, chapter 3: more cuts, more "nice" results).
Rule: a cut of the journal is a hypothesis, not a discovery. You write it down ("does the plan do better in the morning?"), wait until you have enough trades to check it (book 11, chapter 12) and change the plan only then.
What to do with it
- Record every trade at the moment, with the fields above, including a note written before entry.
- Always include the "adherence" field and the reason for every deviation.
- Compute every week mean R, error, losing streak, drawdown (not every day).
- Treat every cut as a hypothesis: note it and wait, do not change the plan.
- Also keep the trades you did not take (skipped signals) with the result they would have had: they are the most useful part.
Procedure in four steps
- Fill in the row at the moment of entry (fields of the plan, note) and at the moment of exit (price, reason, R).
- Mark whether you followed the plan and, if not, which deviation.
- Every week, compute the measures of the plan and the deviations of the week.
- Write down as a hypothesis every pattern you think you see and do not act until you have enough trades.
Exercises
Chapter exercises
1. Which of these journal fields is the most important for measuring the cost of behaviours?
2. In the journal buys make +0.79 ± 0.58 R over 42 trades and sells +0.68 ± 0.64 over 37. How is it read?
3. The 0-7 UTC hour band has +1.66 R over 16 trades, the others less. What do you do?
4. Why do you record the result in R and not only in dollars?
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.