AI Trade Review: One Trade Against Your Own Plan

Because a trade's numbers say what it cost, and only a comparison with the plan says why. Navrya keeps the plan, the session and the strategy's rules beside the trade, so the review reads them together.

The browser collects what the journal links to the trade and sends it when you press the AI analysis button:

• The trade: direction, symbol, entry, stop, targets, size, leverage and result, plus the actual fills if an exchange statement was applied.
• Up to four of its screenshots.
• The session: its thesis, every scenario with the probability you assigned it, and your read of each timeframe.
• The written rules of the linked strategy.
• Your analysis profile: style, focus areas and the concepts you taught it.

The model receives it as data, not as instructions, and the answer is cleaned before it is saved: a concept that is not in your profile is removed.

What comes back is a structured text about that single trade. It opens with a headline and a short summary, marked low, medium or high confidence, and nowhere in it is a buy or sell verdict for the next trade.

• Five probabilities, each with its reason: the target being reached, the scenario playing out, the timeframes agreeing, the plan's quality and the execution's quality.
• The mistakes it found, each with a severity and sorted into nine areas, from entry, stop and target through risk, timing and analysis to psychology, execution and rules.
• Strengths, and suggestions to improve.
• A status for every rule of the linked strategy: followed, broken or unclear.
• Where the trade stands against the session's analyses: aligned, partly aligned, in conflict or unrelated.

Because a review is a snapshot: it records when it was made and from what, and does not update itself. Edit the trade later, or add a new analysis to its session, and the review is flagged as stale, because its text no longer describes the trade or the session as they stand now; run it again for a fresh one.

It reads the trade and the plan together, against the strategy's rules. This example is invented. A trader with an analysis profile built around liquidity opens a London session on one pair, writes one scenario, a sweep and reverse, and gives it a probability. She logs a long with an entry, a stop and a target twice as far from the entry as the stop, and links a strategy with written rules.

1. 01

   After the close she presses AI analysis; the browser sends the numbers, the screenshots, the session with its thesis and scenario, the rules and her profile.

2. 02

   The review returns a headline, a summary and a confidence level, five probabilities with reasons, each rule marked followed, broken or unclear, and the trade's standing against the session's analyses.

3. 03

   A day later she corrects the stop on the logged trade. The review is marked stale, and she runs it again.

The example ends there; the verdict is the review's to give.

It gives no signals, predicts no price, connects to no broker or exchange and places no orders.

• No buy or sell verdict and no suggested entry; the probabilities are estimates, not odds from market data.
• No connection to a broker, exchange or prop firm; a figure read from a statement screenshot must match your own statement.
• It never runs by itself. You press the button on a trade or ask the assistant; until then, logging, the calculator and the account rules work with no AI at all.
• No substitute for the record: an unwritten plan, reason or stop leaves the review just as incomplete.
• Not financial advice. The decisions, and the orders, remain yours.

The review reads what the rest of the journal already holds. The session gives it the thesis, the scenarios and the timeframe reads; a strategy gives it the rules to mark; the analysis profile gives it your lens; and the trade form, in its five steps of Status, Timeframe, What you saw, Emotions and Screenshot, is where the trade's numbers and screenshots are entered. After the close, the result in R and the reflection sit beside it, and the weekly review turns several reviews into one change.

• AI trading journalWhat the AI reads, and what it does not.
• What a trading journal isSetting up, logging in five steps, closing.
• The weekly journal reviewA routine that ends in one change.