Every trader runs experiments all day; almost none of them record the results. A journal is the difference between five years of experience and one year of experience repeated five times — and it is the only trading tool that is literally free while being provably underused.
Why memory fails at this job
Your recall of your own trading is systematically edited. Winners are remembered as skill, losers as bad luck; the impulsive trade that worked becomes "instinct," the plan you abandoned quietly vanishes. Hindsight rewrites what you actually believed at entry. The journal exists because it cannot be renegotiated: what you wrote at 9:14, risking real money, is what you actually thought — and a hundred such records form the only honest dataset about the one variable you can directly improve: you.
What to record
The numbers (mechanical, per trade): date/time, pair, direction, size, entry, stop, target, exit, result in R (multiples of risked amount — the unit that makes trades comparable), setup name from your plan, and session. In R terms, a month is readable at a glance; in dollars it is noise.
The context (one screenshot, two sentences): the chart at entry — marked up, because the chart at exit lies with hindsight — and the reason for the trade in plain words. If the reason takes more than two sentences, that is itself a signal: good setups compress.
The state (the part everyone skips): a one-line honesty note — calm / bored / chasing / revenge / tired — plus whether the trade followed the plan: yes, no, or partially. This column is where the journal earns its keep. P&L is mostly noise trade-to-trade; rule-adherence and emotional state are the signals, and they only become visible in aggregate.
The weekly review: where entries become rules
A journal that is written but never re-read is a diary. Once a week, with markets closed, run three sorts. By setup: which of your named setups actually earn, in R, over the sample — many traders discover one setup funds the account while two others quietly tax it. By adherence: compare the expectancy of plan-following trades against rule-breaking ones; the gap between those two numbers is the current price of your discipline, in R, and seeing it printed changes behaviour faster than any resolution. By state: what do the "bored" and "revenge" trades sum to? For most people the answer funds a nice holiday for the market.
The output of a review is at most ONE change: a new line on the prohibition list ("no trades after two losses — the data says trade three loses 80% of the time") or a demotion/promotion of a setup. One evidenced change a week compounds; five intuited changes a week is thrashing.
Keeping it sustainable
The perfect journal you abandon in nine days loses to the crude one you keep for a year. Sustainable version: a spreadsheet with the number columns, a screenshots folder, and the two honesty fields — five minutes per trade, thirty for the weekly review. Journal the trade when you place it (the reason and state, at minimum) rather than reconstructing at night, because the reconstruction is where the editing creeps back in. And journal the trades you didn't take when the plan said to — skipped valid setups are data about fear exactly the way impulsive entries are data about greed.
Six months of honest records will tell you things no course can: your actual best session, your actual best setup, the actual cost of your worst habit — with your name on the evidence. Most traders never learn these numbers about themselves. That is the entire opportunity.