Trade grading is the score you give each of your trades, usually from 1 to 5, to judge the quality of the process rather than the result. It is the only statistic in this glossary you declare rather than undergo, and that is both its weakness and its whole point.
What the ratio compares
Contents
Every other statistic in a journal is computed from your trades. Trade grading is not: you enter it. It asks you to answer a question the result never poses, did I play this one well, independently of what it returned.
The point of that distinction is that a good trade and a winning trade are not the same thing. A hasty entry that ends in profit is still a hasty entry, and it will repeat. The result makes no difference between the two, and it will reward the second.
Everything rests on one condition: the grade has to be given before knowing the result, or just after the entry. Assigned after the close, it simply follows the P&L, and it then becomes a redundant statistic disguised as a judgment of process.
Average trade grading = mean of the grades given, over graded trades
The scale matters little, 1 to 5 or 1 to 10, as long as it does not change along the way. What matters is comparability over time.
Ungraded trades are excluded. As for everything else, coverage reads next to the average, otherwise the figure describes an unknown sample.
The average alone is unusable. A severe trader and a lenient one produce two incomparable figures, including with themselves six months later.
| Grade crossed with the result | What it says |
|---|---|
| Well graded and winning | The normal case, and the least instructive. The process worked. |
| Well graded and losing | Bad luck. Nothing to fix, and that is exactly what to tell yourself. |
| Poorly graded and losing | The process failed and it showed. The easiest to fix, because the lesson is coherent. |
| Poorly graded and winning | The most dangerous case. A bad habit has just been rewarded. |
| A grade that predicts nothing | Either the grade follows the result, or your process has no link with what you earn. |
The two middle rows contain everything this statistic brings. Well graded and losing is the box no other measure can produce: it says there is nothing to change, which is the hardest conclusion to accept after a loss and often the right one. With no grade given beforehand, a loss always pushes you to fix something.
Poorly graded and winning is the opposite box, and the most expensive in the long run. The result rewards a bad process, memory keeps the result, and the exception becomes a habit. Those trades are few and they are the only ones deserving to be reread one by one.
Well-executed trades pay and badly executed ones cost: the grade is predictive for this trader, which is already information. But it is the eleven trades graded low yet winning that deserve rereading, because they are the ones teaching that you can get away without respecting your process.
It is the only tool in a journal able to make that distinction, and it decides what gets fixed. A run of well-graded losses calls for no change of method; a run of badly graded ones calls for one immediately, and it is not the same one.
Without that separation, a trader fixes something after every bad run, which amounts to changing method at the exact moment it should be held. It is the most ordinary way of never giving a system its chance.
Filter trades graded low and winning. They are rare, and they almost always explain a drift: that is where a stop got moved, a size doubled or an off-plan trade taken, and it worked.
Rereading them once a quarter is enough. It is not a punishment, it is the only way to stop a trade's result teaching the opposite of what the process said.
The central flaw is the grade given after the fact. Assigned once the result is known, it follows it mechanically: a gain becomes a 4, a loss a 2, and crossing with P&L then shows only a manufactured correlation. It is the most frequent flaw, and it makes the statistic not merely useless but misleading.
The second is the absence of a shared scale. Your 3 and another trader's describe nothing in common, and neither does your own 3 six months ago if your severity has changed. This figure compares only with itself, over a period where standards did not move.
The third is looking at the average. It says nothing: not the severity of the grader, not the distribution, not the link with the result. All the value is in the crossing, and an average grade displayed alone on a dashboard is decoration.
The classic mistake. Grading your trades after the close. The grade then follows the result without anyone deciding it: a gain becomes a 4, a loss a 2. Crossing with P&L then shows a fine correlation, which proves nothing except the order in which the two were entered.
| Indicator | Formula | What it measures | When to use it |
|---|---|---|---|
| Trade grading | mean of the grades YOU give | Process quality, declared | Separating bad luck from error |
| Plan followed | compliant trades ÷ total | Compliance, as yes or no | Measuring a rule, not a nuance |
| Win rate | winners ÷ total | The result, without the process | Describing a frequency of gains |
| Tagged emotion | trades with emotion ÷ total | The state, not the quality of the act | Linking a state to a result |
It is the only rule making this statistic usable. Given after the result, the grade follows it and manufactures a correlation that teaches nothing. Given at entry, it records what you thought before knowing, and that is exactly what you want to reread.
Two sentences are enough: what makes a 5, what makes a 1. Without that, your scale drifts with your mood, and a drifting figure does not compare over time, including with yourself.
A handful per quarter. Those are the trades where a bad habit was rewarded, and it is the only place rereading changes anything. The other boxes are read, that one is worked on.
It is the score you give each trade yourself, usually from 1 to 5, to judge the quality of your process rather than the result. A hasty entry ending in profit is still a bad trade, and this grade is the only thing in a journal able to say so.
At entry, or just after, and never after the close. It is the rule that decides everything: a grade given once the result is known follows it mechanically, and crossing with P&L then shows only a manufactured correlation.
To separate bad luck from error. A well-graded loss calls for no correction, which is the hardest conclusion to accept and often the right one. A poorly graded win signals a bad habit that has just been rewarded, and it is the only case deserving to be reread.
It says nothing alone. It measures your severity as much as your execution: a lenient trader will show 4.2 and a demanding one 3.1 with exactly the same trades. That figure compares with nobody, and it compares with yourself only if your standards have not moved.
Better to grade all of them or none over a given period, because a partial grade describes a sample you chose, and people rarely choose at random. If logging is too heavy, one binary question, was the plan followed, filled on every trade beats a five-level grade filled one time in three.
In Tradoshi, trade grading is entered in the Trade Review screen, in stars, next to the free review note and the 'plan followed' field. It is data you supply: nothing computes it, and nothing guesses it from the result.
It becomes useful in the Reports screen, where it exists on both sides of the table. As a metric, under the name average trade grading. And above all as a dimension, which lets you sort any statistic by grade: net P&L by grade is exactly the crossing described above, and it reads without computing anything.
To go deeper on this: read the full guide. Every term is gathered in the glossary.