Keeping a trading journal: what to note, what to skip
Updated on
The short answer
Most journals are abandoned after three weeks, always for the same reason: they are used to copy out by hand what the broker already records, meaning the entry, the exit, the size and the result. That work produces nothing and costs ten minutes per trade, at the worst possible moment. What needs noting is exactly what the machine cannot know: why you took the trade, what would have invalidated it, and above all whether you followed your own plan. That last field alone is worth all the rest of the journal.
What a journal is actually for
It is not there to preserve your results, which are already at your broker and will not vanish. It is there to spot what repeats, which you never see trade by trade.
A mistake made once is an accident. The same mistake made eleven times in two months is a habit, and that kind of thing only appears on rereading. A journal you do not reread is useless, and you only reread what is short.
The consequence is counter-intuitive: a good journal contains less than people think, not more.
What not to note
Everything your platform already records: the instrument, the time, the entry price, the exit price, the size, the result, how long the position lasted.
Copying it creates no new information. It takes time, and that time is demanded at the worst possible moment, right after a loss, when all you want is to close the screen. That is where journals die.
The rule is simple: if the data already exists somewhere, it has no business in your handwritten journal.
The three fields that count
What the machine cannot know comes down to three things, each fitting on one line.
- Why I took this trade: the reason, in one sentence, written before or just after the entry;
- What would have invalidated it: what you would have had to see in order to pass;
- Did I follow my plan: yes or no, with no shading.
The third is the most important of the three, and the one everybody leaves out. Without it you cannot tell a bad result from a bad decision, and that is the whole question.
Decision and result are not the same thing
A market carries randomness in the short term, which produces four situations rather than two.
A good decision that ends well, and a bad decision that ends badly: neither teaches anything, everything went as expected.
A good decision that ends badly is the case you have to learn to accept. You followed your plan, the market decided otherwise, there is nothing to fix. Changing method after that trade is the classic mistake.
A bad decision that ends well is the dangerous case. You moved your stop, doubled your size or took an entry that was not in the plan, and you won. Nothing warns you, and you have just learned to do it again. Only a journal that records the decision separately from the result makes that case visible.
What thirty trades tell you
Thirty lines are enough to reveal recurrences that no intuition provides.
The useful questions are always about concentration. Are my losses grouped on one instrument in particular? On a time of day, say the end of the day when I am tired? On the trades where the “plan followed” box says no?
It is also where you discover the unpleasant and useful things: that the instrument you like best is the one costing you most, or that trades taken within ten minutes of a loss end badly almost every time.
One methodological point: those observations only hold on your trades. The statistics of a channel or of a tool say nothing about what happens on your account, as our page about following a signal channel explains.
Pace changes everything
A journal is sustainable or not depending on how many trades you make, and that is not a question of discipline.
Twenty trades a day cannot be documented, whatever your good intentions. Three trades a week are reread in ten minutes on a Sunday, and it is at that pace that the journal becomes a tool rather than a chore. The link between pace and the horizon you chose is covered on our page about choosing your timeframe.
What AlphaGPT does
The factual part fills itself. Positions taken from the tool are picked up from your account, with their size and what became of them, and the targets of one plan are grouped together rather than counted separately.
So you have nothing to copy out, which removes precisely the friction that kills handwritten journals. What remains is to add what a machine cannot know, meaning the three fields above.
What the tool does not do: it does not judge your decisions, does not tell you whether you trade well, and does not fill in the “plan followed” box for you. Nobody can fill that one in for you, and that is exactly what gives it its value.
Trading carries a risk of capital loss.