The 30-Second Trade Log
When You Enter a Position
Immediately after clicking buy or sell, dictate or type into your LLM:
Log this trade:
TRADE: [BUY/SELL] [TICKER] at ~$[PRICE], [SIZE] shares/contracts
DATE: [today]
THESIS: [1-2 sentences — why right now?]
CATALYST: [What triggers the move? Earnings? Sector rotation? Technical level?]
TIME HORIZON: [Days? Weeks? Months?]
STOP/TARGET: [Where am I wrong? Where do I take profit?]
CONVICTION: [1-10]
RISK I'M MOST WORRIED ABOUT: [One sentence]
Format this as a structured JSON entry and append it to my trade log.
That is it. 30 seconds. You have captured the reasoning at the moment of maximum clarity — when you actually know why you are doing this.
When You Close a Position
Close this trade:
TRADE: [CLOSED] [TICKER] at ~$[PRICE]
RESULT: [+/-$X or +/-X%]
OUTCOME: [WIN/LOSS/SCRATCH]
WHAT HAPPENED: [1-2 sentences — did the thesis play out?]
WHAT I LEARNED: [Was I right for the right reasons? Wrong? Lucky?]
WOULD I TAKE THIS TRADE AGAIN: [Yes/No and why]
The Quarterly Review (This Is Where the Magic Happens)
Every quarter, paste your accumulated trade log into your LLM and use this prompt:
Here is my complete trade log for [Q1/Q2/Q3/Q4]:
[PASTE TRADE LOG]
Analyze my trading behavior and tell me:
1. WIN RATE: Overall and by trade type (long vs short, sector, time horizon)
2. BEST TRADE: Which trade had the best risk-adjusted return? What was the thesis?
3. WORST TRADE: Which trade lost the most relative to my conviction level?
4. PATTERN ANALYSIS:
- Do my high-conviction trades outperform low-conviction? (They should)
- Am I cutting winners too early? (Compare target vs actual exit)
- Am I holding losers too long? (Compare stop vs actual exit)
- Is there a sector or setup where I consistently outperform?
- Is there a sector or setup I should avoid?
5. SIZING ANALYSIS: Am I sizing positions proportional to conviction? Or am I equal-weighting everything?
6. TIMING: Do my trades perform better on certain days, around certain events?
7. ONE THING TO CHANGE: If I could only fix one behavior next quarter, what should it be?
Be brutally honest. I want to improve, not feel good.
What This Actually Produces
I have run this workflow on my own trading for two quarters. Here is what I learned:
Q3 finding: My conviction-7+ trades had a 71% win rate. My conviction-4 trades had a 38% win rate. I was right about what I knew — but I was taking too many low-conviction trades that dragged down overall performance. Fix: minimum conviction of 6 to enter.
Q4 finding: I consistently cut winning trades within 3 days of entry, even when my thesis was "2-4 week hold." My average hold on winners was 2.8 days vs my stated target of 15 days. I was leaving 40%+ of the move on the table. Fix: set a minimum hold period rule.
Neither of these patterns was obvious from looking at my P&L. The AI found them because it could see across 87 trades at once and compare my stated intentions vs my actual behavior.
The Infrastructure
Your trade log can be as simple as a text file or a Google Sheet. The AI does not care about format — it can parse messy notes, structured JSON, or plain English paragraphs.
If you want to get sophisticated, store entries as JSON and build a simple script that feeds them to the LLM quarterly. But start simple. A trade journal that exists beats a perfect one that you never build.
The key principle: capture at the moment of decision, analyze at the moment of review. The AI handles the analysis. You handle the 30-second capture.
What is Coming Next Week
Issue #4: The sector rotation signal hiding in earnings transcripts — how to use AI to read 50 transcripts in an hour and spot the theme everyone else will talk about in two weeks.
AI Finance Brief is written by a team running live algorithmic trading systems. Every workflow is tested on real market data.
Free every week. Subscribe at aifinancebrief.com