CLAUDE JOURNAL OPTIMIZATION PROMPT

You are analysing my real trading history. I have attached the export from my trading platform.

Your job is to find behavioural and statistical patterns in this data that are costing me money, and to tell me plainly when there is nothing there.

HARD RULES - READ THESE FIRST
Do not invent a pattern to give me an answer. If the data does not support a conclusion, say so directly. "There is no meaningful signal here" is a valid and useful answer.
Every claim must come with the actual numbers behind it: the sample size, the win rate, the average result, and how it compares to my overall baseline. A claim without numbers attached is not allowed.
Sample size thresholds. Under 20 trades in a bucket, report it as insufficient data and do not draw a conclusion. 20 to 50 is weak and must be labelled as such. Over 50 you can call it meaningful. Never present a 10-trade pattern as a finding.
Distinguish between what the data shows and what might explain it. You may propose an explanation, but label it clearly as a hypothesis, not as something the data proves.
If a column I would need for an analysis is missing or empty in my export, say which analysis you cannot run and why. Do not estimate the missing values.
Do not soften bad news. If my behaviour is costing me money, say the number.
BASELINE - ESTABLISH THIS FIRST

Before any pattern analysis, calculate and state:

Total number of closed trades and the date range covered
Overall win rate
Average win, average loss, and the ratio between them
Expectancy per trade
Largest win and largest loss
What percentage of total profit comes from the top 5 trades

Everything after this is compared against that baseline.

ANALYSES TO RUN

Day of week Win rate, average result and expectancy for each day. Flag any day that sits meaningfully below or above baseline, with the sample size for each.

Time of day and session Group entries by hour and by session. Report which windows carry my results and which ones drain them.

Behaviour after a loss Compare the trade immediately following a loss against my baseline: position size, time elapsed before entering, stop distance, and result. Then do the same for trades following two and three consecutive losses. If size increases or stop discipline changes after losses, quantify it.

Behaviour after a win Same comparison after a win, and after two and three consecutive wins. Look specifically for size creeping up.

Position sizing consistency How consistent is my size? Identify outlier trades where size deviated significantly from my norm, and report how those outliers performed as a group.

Hold time Compare hold time on winners against losers. If losers are held significantly longer than winners, quantify the gap. This is the standard signature of cutting winners early and hoping on losers.

Instrument breakdown Win rate and expectancy per instrument. Flag anything I am consistently losing on, and anything carrying my results.

Trade frequency Trades per day, and whether high-frequency days perform differently from low-frequency days. Look for overtrading.

Streaks Longest winning and losing streaks. Compare against what would be statistically expected given my win rate, so I can see whether my streaks are normal variance or something else.

Time in the month Whether results shift across the month, particularly toward the end. If I am on funded accounts, payout pressure shows up here.

WHAT NOT TO DO

Do not give me generic trading advice. Do not tell me to journal more, be more disciplined, or manage risk better. I want findings from this data only.

Do not rank findings by how interesting they sound. Rank them by how much money they represent.

OUTPUT FORMAT
Baseline - the numbers listed above, stated plainly.
Findings, strongest first. For each one: what the data shows, the numbers behind it, the sample size, how confident you are and why, and a proposed explanation clearly labelled as a hypothesis.
What the data does not support. Anything you checked and found no meaningful signal in. This section matters as much as the findings.
The single most expensive pattern, with an estimate of what it has cost me across this history in money or in R.