Track Your Trades

Start with the principle everything else hangs on. You cannot fix a leak you never measured. Made a trader error today? Write it down. Then watch the count over the weeks. You might be making that mistake far more often than you feel… or far less.

That is the whole purpose of journaling with error tags… "getting rid of these completely" (May 15, 2025). Tag the same mistake enough times and the tally itself starts to shrink. The number embarrasses you into fixing it.

The raw material of trading statistics is the CSV file your platform exports. George's rule is simple and non-negotiable. Save it from every account, every single time.

Passing an eval on the way to funded? Capture that CSV before the account changes hands. Blowing accounts is even more valuable data… "Every time you blow an account, you should be downloading that CSV file" and running it through analysis (Nov 22, 2025). The blown account tells you exactly where the wheels came off.

Cadence depends on the account. Props get exported per eval, pass or fail. For a cash account, pull the file monthly, weekly if you can. The point is to never lose a session to a forgotten download.

Two months of transaction records paint an honest portrait of who you really are at the desk. Once you see it in black and white, you can massage it… reshape the trader the numbers describe.

A CSV alone cannot tell you why a trade worked. It has no idea whether you took a Battle Plan long, a counter trade, or a 3-minute level. That context lives in your tags, and only in your tags.

Tag every trade with the conditions you want to measure. Battle Plan longs. Battle Plan shorts. 1-minute levels versus 3-minute levels. How well do you do when you wait for each one? Tag them and the answer appears. Enter individual trades and "you will get more data" for your heat map (June 14, 2026).

Some tags matter more than the market realizes. Take counter trades… the ones that fight the side in control. Drop a counter tag on each one and George's question answers itself: "How well are you doing with that specific tag?" (June 14, 2026).

Track your setups too, not just your timing. Give a named setup 10, 20, 30 repetitions and you finally know its true success rate…

Separate your primary trades from your secondary ones while you tag. Primary trades are your bread-and-butter setups. Secondary trades like an IB high, an Asia high, or a London low get their own field so they never muddy your core numbers. Track each by its primary and secondary confluence levels and you build a database you can lean on later, "because you're not going to remember that yourself" (June 29, 2026).

Here's the trap almost every trader walks into. The categories you'll want to filter by three years from now… you have to build them into your journal today. Retrofitting old entries is a genuine pain.

Once the fields exist, filtering becomes trivial. Got 500 journal entries and want only the days before Non-Farm Payroll? Add a condition on the special-day field and the software isolates exactly those sessions. Now you judge your performance on that one condition alone, clean.

Decide what to measure while you're at it. George recommends points, not dollars… "Just stay consistent. I recommend points. I think it works better" (Apr 20, 2025). Points travel across account sizes. Dollars don't.

Be patient with the numbers, too. A single field may look meaningless on any given Tuesday. Put in 200 trading days and the pattern emerges. Statistics need a sample before they speak.

This is where tracking pays off. Once the CSV is loaded and tagged, you filter it down to one condition at a time. Show me just Mondays. Show me all my longs on Monday. The software sorts, and a pattern you never felt in real time snaps into view.

The most common revelation is timing. Most traders are excellent in one part of the day and ruinous in another. You bank a clean morning, then hand it all back over New York lunch. Maybe the afternoon does it. Maybe you go negative in the evening and spend the RTH session digging out.

Direction is the next slice, and George's own data drove a real pivot. He held shorts too long. Same count of 10-point winners as his longs, far more that ran into 10-point losers. Reading that number is "one reason I have become more long focused" (Apr 30, 2025). The data made the decision, so the decision stuck. A few of the highest-value slices:

One more slice worth logging by hand: heat. Record how far price moved against your entry on every trade. Two points against a break level, one point against another. Measure it every time and you build reliable data on how much room each setup demands. Split your AM versus PM results as well (Apr 20, 2025)… the AM-PM gap alone reshapes many trading days.

Before you trust any statistic, scrub the outliers. A forgotten order left in your DOM. A freak win or loss that has nothing to do with your process. Delete those individually so they don't distort the true picture… "I don't want that in my metrics" (May 26, 2025). Hate your Sunday trades? Pull them all. Your numbers should describe how you actually trade.

Now hand the clean file to an analyst. That analyst can be AI. Drop your CSV into the Chart Buddy trade analyzer, or straight into ChatGPT, and ask it to break down your metrics…

The trade analytics finds which days and times you trade best and worst automatically… "it does all kind of analysis for you" from the daily journal software (Jan 25, 2026). Journaling in markdown? Download it and feed it to an AI coach with the supplied prompts: patterns, discipline, setup analysis, blind spots. The machine spots the leaks you've gone blind to.

Trust, then verify. AI is a powerful analyst and an imperfect one. George caught ChatGPT returning a wrong figure mid-session… "3128, that is not correct" (Nov 24, 2024). His fix is a habit worth stealing.

Sanity-check every output against a number you already know. ES averaging roughly 50 points a day. Your own prior read on a stat. When AI matches the benchmark, trust the rest. And feed it plenty… squish the chart, export as many days as it will give you, and download several times to make results statistically significant.