Grade each trade's decision before you see its result, then look for lucky wins

How it works

1. During the week, journal each trade as usual: the plan, what you were thinking at entry and at exit, and a screenshot. Keep the result in its own column.
2. At the weekly review, hide the result column (or ask the agent to show you the entries without it) and grade each trade's process from A to D against your written rules: planned setup, planned size, stop placed, exit rule followed.
3. Write one line for each grade saying which rule it rested on. You grade; the agent can point out where your note contradicts your grade, nothing more.
4. Show the results again. The agent sorts trades into four boxes: good process and good result, good process and bad result, poor process and good result, poor process and bad result.
5. Spend most of the review on poor-process wins (lucky wins, the habits being rewarded) and good-process losses (the cost of following the plan, worth accepting). Write one line for each.
6. Track over time what share of your wins came from poor process. If it grows, your rules are being bent and the results are hiding it.

Copy-paste prompt

You help with my weekly process review. You do not give trading advice and you never suggest trades, markets or
securities.

My rules: {{rules}}
This week's trades: {{trades}} (id, plan, notes at entry and exit, my process grade A to D, the rule I based the grade on, result in R)

1. For each trade, check whether my notes contradict my grade (for example a grade of A where the notes say the stop
   was moved). Quote the note and the rule. Do not regrade it yourself.
2. Sort trades into four boxes: good process (A or B) or poor process (C or D), against good result (above 0R) or bad result.
3. List the poor-process wins and the good-process losses with my notes on each, and ask me one question about each.
4. Give the share of wins that came from poor process this week, next to the earlier weekly shares in {{history}}.
Do not write to my journal; I will copy anything I keep.

Guardrails

• You grade the process before you see the result. The agent only flags where your notes contradict your grade.
• Grades rest on your written rules, cited per trade.
• No opinions on securities or setups; the review covers how you followed your process.
• Results are in R or percent of account from your own journal, not currency amounts shared with the assistant.
• The agent writes nothing; you copy the lines you keep.

What can go wrong

The blind grade is only blind if you grade before you remember the result, and with recent trades you usually do remember. Grade the oldest trades first, or grade daily before results are known, and treat the weekly pass as a check. The other failure is grading generously, so that everything lands in the good-process boxes. That's what the contradiction check is for. A poor-process share near zero for months is more likely to mean lenient grading than perfect discipline.

Source and references