How to grade your own trades (and check whether your A setups actually win)

Grading your own trades is one of the oldest pieces of advice in trading, and it is usually taught badly. The instruction is "grade each trade A to F," and then nothing — no definition of what earns an A, and no step where you go back and find out whether the letters meant anything. A grade you never audit is a diary entry. It feels like process and measures nothing.

This post is the version that closes the loop: what the letters should be attached to, how to assign one in about five seconds, and the check that makes the whole exercise worth doing.

Grade the decision, not the outcome

The single mistake that makes grading useless is grading on P&L. If a winner is automatically an A and a loser is automatically an F, the grade carries no information the P&L column didn't already have, and you've spent five seconds per trade re-typing your results.

A good trade can lose money. A bad trade can make money. Over any short sample, a disciplined trader taking correct setups will have red weeks, and an undisciplined one will have green ones. Grading exists to separate those two things, which means the grade has to be assigned on what you controlled: whether the setup qualified, whether you sized it right, whether you honoured the stop, whether you exited where you planned to, and whether you were in a fit state to be trading at all.

The practical test: you should be able to assign the grade the moment you enter the trade, before you know the outcome. If your grading method can't do that, it's outcome grading wearing a costume.

Define the letters with a checklist, not a feeling

"A-quality" means nothing until it's a list of boxes. Write five questions you can answer yes/no about any trade you take, then let the count of misses set the letter:

• A — every box ticked. The setup qualified, size was right, stop honoured, exit per plan, head clear.
• B — one miss. Usually a real trade with one piece of slippage in the execution.
• C — two or more misses. The setup may still have been fine; you didn't trade it the way you said you would.
• D / F — you broke a rule you had written down before the session. Sized up after a loss, moved a stop, traded outside your hours, took something that wasn't on the watchlist. The P&L is irrelevant here: an F that made money is the most expensive kind, because it pays you for the behaviour that eventually empties the account.

Five questions is the right number because it fits in your head and it takes seconds. The five worth starting with, if you don't already have your own: did I follow my plan, did I size correctly, did I honour my stop, did I exit where I planned, and was I emotionally clean. Those five cover the large majority of self-inflicted damage in a discretionary account.

The step everyone skips: audit the grades

Here is the part that turns grading from a ritual into a measurement. Once you have a hundred graded trades, group them by letter and compute average R-multiple and win rate for each group.

You are looking for a monotonic curve — A above B, B above C, and D/F clearly negative. That is the result that tells you the checklist is capturing something real about your edge.

When the curve comes out flat, that is a finding, not a failure. It means your checklist is measuring effort rather than edge — the questions feel virtuous but don't discriminate between trades that work and trades that don't. Replace the questions that don't move the outcome with ones drawn from your actual losers. When the curve comes out inverted, with C trades outperforming A trades, the usual cause is that your "A setup" definition is too tight and you're only taking it in conditions where it's already crowded. Both of those are useful, and neither is visible without the audit.

The per-question rollup is where the specific fix lives

The letter tells you a trade was off. The five yes/no answers tell you which part. Count the no's per question across your history and one of them will usually dominate — a stop-honouring problem and a sizing problem are both C grades and have completely different fixes.

This is the number worth putting on a sticky note: the single question you answer "no" to most often. It is, empirically for you, your most expensive habit. Most traders can name the habit; very few can name it with a count attached, and the count is what makes it stop feeling like a personality trait and start feeling like a fixable rate.

How long before the numbers mean anything

Grades over 20 trades are noise, in exactly the way a win rate over 20 trades is noise — one outlier winner in the C bucket will invert the whole curve. A hundred graded trades is where the comparison starts to carry information, and that's an order of magnitude rather than a precise threshold.

The corollary matters more than it sounds: don't act on the first flat curve you see. Grading is a habit with a long payback period, and the traders who abandon it usually do so at trade 30, right before the sample gets large enough to say anything.

Three ways grading goes wrong

• Grading in a batch on Sunday. By then you remember the outcome and not the decision, and every grade quietly bends toward the P&L. Grade at the close of the trade, or at the end of the session at the latest.
• Grading generously. If 90% of your trades are A's, the scale has collapsed and it can't discriminate. A distribution with nothing below B usually means the checklist questions are too easy to answer yes to.
• Grading only the losers. The bias is understandable and it destroys the comparison — you cannot compute average R by grade if the A bucket is missing every winner you didn't bother to review.

How TradeFlow Quantum handles this

Each trade carries a self-grade (A through F) and a five-question process review — followed plan, sized correctly, honoured stop, exited at plan, emotionally clean — filled in from the trade's own page rather than a separate document you'd have to keep in sync.

The audit step is the /analytics/grades view: it groups your graded trades by letter and reports count, average P&L, average R-multiple and win rate for each, plus a per-question rollup counting yes / no / n/a across your history. That's the flat-or-monotonic curve described above, computed from your own trades.

One thing to be clear about, because it's a real distinction and not a modesty note: the app does not grade your trades. You assign the letters; the app aggregates what you assigned and shows you what happened afterwards. "Of the trades you graded A, here's what followed" is a very different claim from "these were good trades," and only the first one is honest about where the judgment came from.

Honest disqualifier

If you trade a fully systematic strategy with no discretion at execution, skip this entirely. Your rules are the discipline, your grades would be A on every trade by construction, and the audit would return a flat curve for a reason that has nothing to do with your edge. Grading earns its five seconds for discretionary traders making real-time judgment calls — which is also who the rest of this site is written for.