The problem with reviewing one trade
A trade is rarely a mistake on its own. The third breakout chased on the same ticker in a week, or a full-size entry straight after two losses, reads very differently from a one-off. An AI that sees only the trade in front of it cannot tell those apart — and if it tries, it invents a count.
Illustrative sample data — not actual or achievable results.
What the autopsy sees now
Before the AI is asked anything, the app reads your earlier trades on the same account and writes down what it finds, as finished sentences:
- where this trade falls in the 7 days up to its entry
- how many trades on the same symbol this week, and how many of them lost
- how many trades on the same Blueprint this week, and how many of them won
- each mistake tag on the trade, counted across the week
- the trade before it and its result, and any winning or losing streak of 2 or more
- its size against the median of your previous 20 sized trades, once there are at least 5 to compare
A count the record does not state is removed
The AI is told to use those facts and nothing else. Then the app checks its reply: a sentence claiming a number of trades, wins or losses that is not in the facts is dropped before you see it, and before it is saved. Only earlier trades on the same account are counted, and planned trades are left out.
The rest of the autopsy is unchanged: entry quality, how the exit followed the plan, the pattern, two or three things to work on, written from your own notes, tags and rating. It never predicts a price or calls a buy or a sell.
What it costs
Each autopsy uses one AI action from your plan's allowance (3 on Free, 30 on Essential, 200 on Pro each billing period). Opening a saved one again is free; Regenerate uses another action, and a failed one uses nothing.
Open it: Journal → any closed trade → AI autopsy
Open in the app