5 things to log
after every session.

Log your actual net P&L after commission. Not gross. Not "roughly." The exact number.

This sounds obvious, but a surprising number of traders log approximate figures or avoid logging losing days entirely. Both habits make the data useless. The AI can't detect patterns in partial data, and you can't improve what you won't measure.

What Claude looks for: P&L relative to your recent average. A $200 loss means nothing in isolation. A $200 loss on a day where your average is +$400 tells Claude something is off — your size, your setups, or your execution. The trend across sessions matters more than any single number.

Log how many trades you took and how many were winners. If you took 8 trades and 5 won, that's 62.5% — log it as 62 or 63.

Most traders either skip this or treat it as a grade. It's neither. A 40% win rate with a genuine 3:1 true RRR is more profitable than a 70% win rate at 0.8:1. Win rate only means something in combination with your average risk-reward.

What Claude looks for: The gap between your expected win rate (from the calculator) and your actual win rate. If you're running the calculator at 65% but actually hitting 48% in your log, Claude will flag that your real edge is weaker than your setup suggests — or that execution is costing you winners you should be taking.

It also looks for consistency. A 65% win rate over 20 sessions is a very different signal than 85% one week and 40% the next. Variance is information too.

Log how long you were actually in the market. Not how long you were at your desk — how long you were actively trading.

This field catches one of the most expensive habits in futures trading: overstaying. A trader who makes $800 in the first 90 minutes then gives back $400 over the next two hours has a session length problem, not a strategy problem.

What Claude looks for: The relationship between session length and outcome. Claude will notice if your best sessions are consistently shorter than your worst ones — which is true for most traders and almost nobody admits it. It will also notice if your losing sessions are getting longer over time, which is a classic sign of revenge trading or trying to "make back" losses.

Log your account balance at the end of the session. Yes, every session.

This lets MICROEDGE calculate your equity curve, your drawdown from peak, and your progress toward doubling. None of that works if your account size is stale.

What Claude looks for: Drawdown patterns. If your account drops more than 5% in a session, Claude will call that out directly — not as a moral judgment but as a process flag. Prop firm traders especially need this: your account size logging is what tells the AI whether you're approaching your daily loss limit or getting dangerously close to a drawdown that would end a challenge.

It also tracks growth rate. If you've been trading for 60 days and your account is flat, that's data. If it's growing at 2% per month consistently, that's also data — and Claude's coaching will reflect which situation you're in.

This is the field most people get wrong. Either they write nothing, or they write a novel justifying every trade. Neither is useful.

The notes field should answer one question: what actually happened today that's worth knowing tomorrow?

The bad example has nothing specific in it. The good example has a behavior (trading after target), a number ($380), and a pattern (chasing). That's what Claude works with.

What Claude looks for: Specific behaviors. Words like "chased," "re-entered," "moved my stop," "got impatient," "sized up," "held too long," "cut too early" — these are the raw material of useful coaching. Generic descriptions of market conditions are not.

Here's an example of how the same session notes produce very different coaching responses:

The difference is specificity. Give Claude something specific to work with and it will give you something specific back.