Trading Ledger

Overview

A journaling skill in the tradition of the Market Wizards interviews: a written record of every trade's decision process, reviewed on a schedule. The user reports a trade in plain language — "bought 500 NVDA at 135, stop at 128, betting the post-earnings dip fills" — and the agent writes ticker, size, and price to the user's own Notion database plus the part every spreadsheet journal loses: the thesis, the plan, and the emotion. If no reason is stated, it asks on the spot, because entry reasons decay overnight. Reviews grade decision quality against the user's own plan — a per-plan loss scores better than a lucky win. Core contract: never fabricate when unsure; mark To-confirm and batch-ask.

When to Use This Skill

• Use when the user reports a trade fill ("bought 500 NVDA at 135", "closed my TSLA position", "opened 2 ES contracts short")
• Use when the user says "log a trade" or "trading ledger", or "tidy up my trading ledger"
• Use when the user asks to "review my trades"

How It Works

Step 1: Confirm the database (first use per session)

Use Notion search to find candidate databases (type database, not pages) whose title contains "trading-ledger". Before any query or create-pages call, show the user the candidate title and data_source_id (collection://... UUID) and ask them to confirm the exact database for this session. A single title match is not sufficient confirmation. If the connector returns owner or schema metadata, show it as an additional identity check. Use only the user-confirmed ID for the remainder of the session; do not re-run fuzzy selection after a confirmation.

The companion Notion template (free, linked in the source repo) ships this schema — select values are a controlled enum, copy them exactly:

• Entry (title); Ticker (text — strikes/expiries here: NVDA / NVDA 0620C150 / ESU6)
• Market (select): US Stocks / US Options / US Futures / A-Shares / HK Stocks / CN Futures / Crypto / Other
• Direction (select): Long / Short · Size (text, with units)
• Entry Price / Exit Price (number) · Entry Date / Exit Date (date — expand to "date:Entry Date:start": "YYYY-MM-DD"; a bare value fails with HTTP 400)
• Thesis (text) — the soul of the journal; if missing, ask on the spot
• Plan (text) — stop / target / contingency; ask if missing
• Emotion (select): Calm / FOMO / Panic / Revenge / Boredom / Overconfidence — tag only what the user admits or what is plain in their words; don't diagnose
• Execution (select): Per plan / Early exit / Delayed stop / Impulse / Unplanned add — filled at close
• P&L (number, realized) · Status (select): Open / Closed / To-confirm / Reviewed · Review (text) · Notes (text)

Step 2: Opening a trade

Parse → create a row with Status=Open. A missing thesis is the one field worth interrupting for — ask immediately. Anything else uncertain: record what you have, put the question in Notes, mark To-confirm.

Step 3: Closing or adjusting

Find that ticker's Status=Open row → fill Exit Price / Exit Date / P&L / Execution → set Closed. Grade Execution against the user's own Plan: stopped where planned = Per plan; ran before the target = Early exit; held through the stop = Delayed stop. No matching open row → create one marked To-confirm and ask whether the entry was never logged.

Step 4: Batch reconcile and review

• "tidy up my trading ledger" → query Status=To-confirm rows → collect all open questions into one message.
• "review my trades" → query recent Closed + all Open rows. For each closed trade, three questions: Did the thesis play out? (wrong thesis + profit = luck — say so). How was the execution? (a per-plan loss is a good trade). What share of trades were emotion-tagged? Write conclusions into Review, move Closed → Reviewed, and for every open position ask: does the entry thesis still hold today?

Parsing rules: infer Market from symbol and context (ambiguous → ask); bought put = Short exposure + note it is a long put; record only given prices/sizes — never look up market prices to fill gaps; "today" = the user's local trading date — confirm before writing.

Examples

Example 1: Opening with a missing plan

User:  picked up some TSM July calls this morning, same thesis as before earnings
Agent: Logged ✅ TSM July calls · Long · Open

…