What to Actually Record in a Trading Journal

The short answer

A trading journal is not a diary. Its job is to hold information that will not exist later, so that questions you cannot answer today become answerable in six months.

That makes the timing more important than the content. Anything written after you know the outcome is a reconstruction — not dishonest, just unavoidably shaped by what happened. The fields that matter are the ones recorded before the result is known, and there are only three of them.

1. Why this trade, in one sentence.
2. The level or condition that would make it wrong.
3. The size, and the arithmetic that produced it.

Everything else — entry, exit, outcome, costs — your platform already records, and will export. If your journal is mostly those, you are duplicating a statement and calling it review.

Why "before" is the whole argument

Every piece of research this publication cites shares one property: the data was collected before anyone knew how it would turn out. Barber and Odean's household records, the complete Taiwanese market history, the daily emotional-state surveys of eighty day traders — all of them are contemporaneous. That is what makes them evidence rather than recollection.

Your own trading is the same problem at a smaller scale, and it has the same solution. A record made at the time can tell you whether you followed your rule. A record made afterwards tells you what you now think you were doing, and the research on self-assessment is not encouraging about that. Work on confidence separates overestimation of your own performance, overplacement relative to others, and overprecision — being too certain your estimate is close — and finds the three close to uncorrelated and two of them weakly measured. There is no version of introspection that substitutes for a timestamp. We go into this in telling confidence from overconfidence.

The fields

Before the trade

The reason, in one sentence. Written so that it can later be found to have held or not. "Breakout from a three-week range with volume" is a sentence that can fail. "Looked good" is not. This field is also a filter in its own right: a trade you cannot summarise in a sentence is usually a trade you have not thought through.

Where the idea is wrong. A level, a condition, a time. Decided from the market rather than from the position size — the reasoning is in stop-loss discipline. Recording it is what makes it possible, months later, to count how often you honoured it.

The size, and how you got it. Not just the number. The amount a mistake was allowed to cost and the distance to invalidation, so the division is visible. This is the field that exposes the commonest sizing fault, described in position sizing when it is uncomfortable: if your size is near-constant while your losses vary widely, you have been choosing size first and fitting the stop to it.

Optionally, and usefully: the state you were in. Not a mood essay — a tick box. Was a position still open elsewhere? Had the previous trade lost? Is this the first trade of the session? Those three are enough, and they correspond to conditions the research actually flags: risk-taking rose after an unrealised loss in the experimental work, which is a condition you can record in one character.

After the trade

What you did, and when. Timestamps, exported.

Whether the written rule was followed. A yes or a no, not a paragraph. This is the only after-the-fact field with real analytical value, because it is the one that pairs with the before-fields.

The outcome. Last, and least. It is the field everyone starts with and the one that answers fewest questions, because a good decision can lose and a bad one can win. Over a few hundred trades the outcome column becomes informative. Over ten it tells you almost nothing — see what a losing run does to judgement.

The questions a journal should be able to answer

Design the record backwards from the questions. These are the ones worth being able to answer, and each requires a field above.

Question	Field it needs
What proportion of my trades met my own criteria?	The reason, written before
How often do I move a stop, and in which direction?	The level, written before
Is my size an output or an input?	The derivation, written before
How often do I trade within an hour of a loss?	Timestamps plus the state tick box
What is my longest run of losses to date?	The outcome column, over time
What share of my result is transaction costs?	Costs, exported

That last one is not a formality. Across the whole Taiwanese market between 1995 and 1999, individual investors' losses broke down as 27 per cent trading losses, 32 per cent commissions, 34 per cent transaction taxes and 7 per cent market-timing losses — roughly two-thirds of the damage was the cost of transacting rather than being wrong. If you have never calculated that ratio for yourself, you are missing the component that mattered most in the largest study of the question. The frequency side of it is in why people trade too often.

What not to put in it

Long narrative entries. They feel productive and they are the least falsifiable thing you can write. If a note cannot be turned into a count, it will not survive contact with a review.

Screenshots as the primary record. A chart image shows what happened. It does not show what you decided, which is the thing you are trying to capture.

Emotion scores you cannot act on. A one-to-ten anxiety rating produces a column nobody ever uses. The tick boxes above are better because each one corresponds to a condition with a documented association, and each one can be cross-tabulated against whether you followed your rule.

Anything you will not maintain. A three-field record kept for a year beats a twenty-field template abandoned in March. This is the same argument as the one about plans: the research on if-then planning found its advantage concentrated in actions that are hard to initiate, and a journal with twenty fields is hard to initiate every single time.

How to review it

Not after every trade. The sample is one and the conclusions will be noise.

Monthly, or every fifty trades, with the questions above in front of you and the answers computed rather than remembered. The point of the review is to produce numbers, and then to decide — in advance of the next period, when nothing is at stake — what if anything changes. Writing the change as an if-then rule is what gives the next month's record something to test, as described in what a plan has to decide in advance.

A journal does not improve a method. It is the only instrument you have for finding out whether you are running the method you believe you are running, which is a different and more basic question, and one that the evidence on trader behaviour suggests most people answer wrongly.

Nothing here is advice. Trading leveraged products carries a high risk of losing money quickly.