Trading Mistake Taxonomy: Frequency and Severity Tracking

Most traders chase their most frequent mistake. That is often wrong. The error that costs the most money is rarely the most common. A two-axis taxonomy — frequency and severity — tells you which fix actually moves P&L.

The two axes

Frequency = how often the error occurs per 50 trades. Severity = average R-cost per occurrence (how much it cost relative to your standard risk). Plot every error type on this grid and fix in a specific order.

The taxonomy categories

• Execution: chase, hesitate, wrong size, wrong stop, missed entry. Typically high frequency, low severity (0.2–0.5R each).
• Risk: oversize, remove stop, average down, ignore daily limit. Low frequency, high severity (1–3R each, sometimes account-threatening).
• Plan: B-grade setup, forced trade, against-trend, ignored macro filter. Medium frequency, medium severity (0.5–1.5R).
• Psychological: revenge, FOMO, early profit-taking, stop-widening. Medium frequency, high severity (1–2R, and often clusters into multi-loss streaks).

The priority matrix

Sort errors into four quadrants and fix them in this order:

1. High frequency × high severity (the killers). Usually a psychological cluster — revenge trading or stop-widening after a loss. Fix first with a structural intervention: mandatory 15-minute break after two losses, automatic size reduction after any violation. One fix here can change monthly P&L more than any strategy tweak.
2. Low frequency × high severity (the landmines). Oversizing or removing stops. Rare but catastrophic. Fix with a hard rule plus a circuit breaker: any single trade risk above 1.5% closes the account for the day.
3. High frequency × low severity (the bleed). Chase entries and hesitation. Fix with a pre-trade checklist and limit-order-only execution. Bleed is slow but compounds.
4. Low frequency × low severity (the noise). Ignore until it migrates up.

Tracking the cost, not the count

Tag every error with its R-cost, not just a tally. A revenge trade that cost 2R counts more than ten 0.1R chase entries. Sort your error ledger by total R-cost over 50 trades and attack the top item only. Re-rank every 50 trades — the ranking shifts as you fix the top leak.

The fix loop

For the top error: identify the trigger, design one structural fix (not a willpower fix), run it for 20 trades, measure whether the error's R-cost dropped. If not, escalate the fix. Willpower-based fixes fail because they rely on the same stressed brain that caused the error.

The bottom line

Classify errors by frequency and severity, then fix by total R-cost — not by count. The high-frequency, high-severity psychological cluster is usually the top P&L leak and needs a structural, not willpower, fix. Track R-cost per error, re-rank every 50 trades, and attack only the top item until it drops.