What to Actually Record in a Trading Journal

Most trading journals are abandoned within six weeks. The usual explanation is discipline. The real reason is that the journal never answered a question, so maintaining it stopped feeling like work that paid.

A journal earns its keep when you can query it. That means the fields you record have to be the ones you will later want to group by, filter on, and compare. Everything else is decoration.

The fields that produce answers#

These are the columns that let you ask useful questions later. If a field cannot appear in the sentence "my performance is different when ___", it probably does not belong.

Instrument and market. Obvious, but record the market type too (forex, crypto, stocks, futures). Traders who work across markets almost always find one of them is subsidising the others.

Setup or strategy name. The single highest-value field in the whole journal, and the one most often skipped. Without it you cannot separate a working strategy from a broken one, and your aggregate statistics will describe a blend of systems you do not actually trade as a blend.

Date and time of entry. Time-of-day performance is one of the most reliable patterns in retail trading records. Sessions matter, and the first hour after an open is a different market from the fourth.

Direction. Long versus short performance frequently diverges sharply, particularly for traders who learned in a one-directional market.

Planned risk, in percent and in currency. Not just the stop price — the actual amount at risk when the trade opened. Without this you cannot calculate R-multiples, and without R-multiples you cannot compare trades of different sizes.

Entry, stop, and target prices. The stop must be the original one, recorded before the outcome was known. A journal where stops are backfilled after the fact is worse than no journal, because it produces confident wrong conclusions.

Outcome and P&L. The result, plus enough detail to know whether you exited at target, at stop, or discretionarily.

Account balance at entry. Lets you reconstruct risk percentage accurately as the account grows, and makes drawdown measurable.

The fields that feel useful and are not#

Long free-text notes on every trade. They do not aggregate. Three hundred paragraphs of reflection cannot be grouped, filtered, or counted. Keep notes short and reserve them for trades that were genuinely unusual.

Emotional state on a 1–10 scale. This sounds rigorous and almost never is. Self-reported scores drift, and the number recorded after a win is not measured on the same scale as one recorded after a loss. A binary "did I follow my plan: yes or no" is cruder and far more useful.

Screenshots of every trade. Valuable for a handful of instructive trades, a burden for all of them. The marginal chart is not worth the friction that stops you logging at all.

Indicator values at entry. Unless you are systematically testing a specific indicator threshold, these fill columns and answer nothing.

The checklist that matters more than the notes#

The most useful qualitative field in any journal is not a description. It is a small set of yes/no questions asked the same way every time:

• Did this setup meet all my entry criteria?
• Was the risk within my planned limit?
• Was the stop placed before entry and left alone?
• Did I exit according to plan, or discretionarily?

Four booleans. They aggregate perfectly, and the resulting statistic — your plan-adherence rate — usually correlates with performance more strongly than any market variable in the journal.

If your plan-adherence rate is 60%, no amount of strategy refinement will help, because you are not trading the strategy you are refining.

Reviewing, which is the part people skip#

Recording is not journaling. The journal only pays when you read it back, and the review needs a cadence and a fixed set of questions.

Weekly, ten minutes. How many trades, how many broke a rule, and what was the worst rule break?

Monthly, half an hour. Expectancy per setup. Performance by time of day and by direction. Any loss worse than −1R, and why.

Quarterly. Whether each setup still has a positive expectancy on a rolling window, and whether anything should be retired.

The point of the fixed questions is to prevent the review from becoming a narrative. Left to ourselves we explain our results; the questions force us to count them.

Spreadsheet, Notion, or a dedicated tool#

A spreadsheet is a perfectly good start and costs nothing. It breaks down at the point where you want grouped statistics across several dimensions without rebuilding pivot tables, and where manual entry starts costing enough time that you skip trades — which quietly biases the whole dataset, because the trades people skip logging are not a random sample.

Notion is pleasant to write in and poor at aggregation, which is the opposite of what a journal needs.

Tradexy exists for the stage after that: structured fields, per-setup and per-session statistics computed for you, checklists that aggregate, and risk metrics derived from data you already entered. If you trade a specific market, the forex, crypto, and prop firm pages cover what that looks like in practice.

Whichever you choose, the test is the same: in three months, can you answer "which of my setups actually makes money, and when?" If the answer is no, the fields are wrong.