Mental track
Journaling and review that changes behaviour
A journal that records how you felt is a diary. A journal that records whether you followed your own rules is evidence, and evidence is what lets you improve something other than your mood.
The fields that do the work
The value of a log comes from the columns you commit to before you know which ones will matter. These are the ones that answer the questions you will actually have in a month.
| Field | Answers the question |
|---|---|
| Setup name | Which of my setups is actually paying for the others |
| Plan followed: yes or no | Is my problem the rules or my execution of them |
| If no, what deviated | Which specific rule I break under pressure |
| Risk in currency | Whether my sizing is what my plan says it is |
| Stop distance planned and actual | Whether stops are being managed or moved |
| Exit reason | Whether trades end by rule or by impulse |
| Instrument and session time | Whether some hours are consistently worse |
| State before entry, one word | Whether a state precedes bad decisions |
Note which field is deliberately binary. "Plan followed" is yes or no, not a rating out of ten. A scale invites negotiation, and the whole purpose of the field is to produce an unambiguous number: your plan adherence rate.
Adherence rate is the metric to watch first
Count your trades for the week. Of those, how many followed the plan? That ratio is the first number to track, ahead of profit and loss, and for a specific reason. If the rules are sound but adherence is 60%, then no amount of work on entries will help, because a third of your results have nothing to do with your strategy. The fix is behavioural. Conversely, if adherence is 95% and results are poor over a large sample, now you have a properly framed question about the rules.
It is the only number you can improve immediately, it is not affected by market conditions, and it is measured entirely by you. Profit and loss over a small sample is mostly noise, so a week spent optimising against it is a week spent optimising against randomness. This is the same sample size problem described in Testing a strategy without fooling yourself.
A review cadence that survives a bad month
- Daily, ten minutes, decisions only. Fill the fields while the session is fresh. Read nothing into the profit column. The daily pass is for recording, not for judging.
- Weekly, thirty minutes. Compute your adherence rate for the week, look at which deviations repeated, and pick one change. One. Then define how long you will run it before judging it.
- Monthly, unchanged. Compare your drawdown against what your testing predicted, and check whether the losing runs you experienced are inside the expected range.17
Why not to review profit and loss intraday
This is not a preference, it is the practical consequence of two findings already covered on this site. People are "distinctly more sensitive to losses than to gains", and the effect grows with how frequently results are evaluated.15 Separately, day traders whose emotional reaction to gains and losses was more intense, in either direction, performed significantly worse.14
Checking profit and loss between trades is a way of evaluating your results many times a day, which is the mechanism both findings point at. The journal is where outcomes get looked at properly, on a schedule, with enough context to mean something.
One change at a time, with a period attached
The most common way a review process destroys value is by generating too many changes. Three adjustments a week, judged after two days each, is not iteration. It is a search for a backtest that fits the last twenty trades, which is exactly the overfitting problem in the testing guide.
Attach a period and a criterion to every change:
- Write the change as a rule, in the plan, with today's date.
- Write the number of trades it will run for before you judge it. Fifty trades is a reasonable floor, given what the sample size table in the testing guide shows.
- Write what would count as failure.
- Change nothing else until the period is complete.
Experience is not automatically the teacher
It would be comforting to think that a few hundred trades teaches the lesson by itself. The evidence is against it. In the Taiwanese day trading study, unprofitable traders were more likely to quit than profitable ones, and yet the aggregate performance of day traders was negative, the vast majority were unprofitable, and many persisted despite an extensive experience of losses.13
Time in the market, by itself, did not convert into skill for most of those traders. What separates a useful review from simply accumulating hours is that the review has fields, a number attached to adherence, and a limit on how much is allowed to change at once.
Fill the log fields. Tick whether each trade followed the plan. Note the one deviation that repeated. Close the platform. That is the whole of it, and done consistently it will tell you more about your trading than any indicator you could add.
Sources
The number next to each claim above is a stable source id. The same number means the same source on every page of this site.
- 13Do Day Traders Rationally Learn About Their Ability?Barber, Lee, Liu and Odean, UC Davis and UC Berkeley
- 14Fear and Greed in Financial Markets: A Clinical Study of Day-TradersLo, Repin and Steenbarger, American Economic Review 95(2), NBER Working Paper 11243
- 15Myopic Loss Aversion and the Equity Premium PuzzleBenartzi and Thaler, The Quarterly Journal of Economics 110(1), NBER Working Paper 4369
- 17The Deflated Sharpe Ratio: Correcting for Selection Bias, Backtest Overfitting and Non-NormalityBailey and Lopez de Prado, The Journal of Portfolio Management 40(5)