After a losing session, most traders reach the same dead end: they know something went wrong, but they cannot name it precisely enough to fix it. "I made a bad trade" is a feeling, not a diagnosis. Without a diagnosis, the review produces no actionable change — only a vague resolution to "be more careful." This article gives you a classification framework that turns that vague feeling into a specific category, and a counting method that reveals which category is costing you the most over time.

By the end, you will be able to assign one of five labels — or a deliberate non-label — to any closed decision, and tally them across a sample of decisions to identify your dominant failure type. The fix differs sharply depending on which type leads the count.

Why "I Made a Mistake" Is Too Vague to Fix

A decision error and a bad outcome are not the same thing. A decision can be sound and still lose — markets are probabilistic, and no process wins every time. Conversely, a decision can be badly flawed and still produce a gain, which is in some ways the more dangerous outcome because it rewards the wrong behavior.

When you conflate the error with the outcome, you end up fixing the wrong thing. A trader who reviews a loss and concludes "I need a better entry signal" may have had a perfectly valid entry — the actual failure may have been in how much risk was taken, or in an emotional state that caused premature exit. Studying more charts will not help. A trader who reviews a win and assumes no error occurred may have been running an undisciplined process that happened to work once. The win obscures the real problem.

The taxonomy below forces a separation between what happened to the price and what happened to the decision. Those are different questions, and they require different answers.

The Five Categories

Every decision error belongs to one of five categories. Each has a distinct cause and a distinct fix. Mixing them together — treating all errors as the same kind of problem — is how correction efforts fail.

Analysis error. The read of the situation was wrong before any action was taken. The model, the interpretation of context, or the identification of conditions was inaccurate. This is a knowledge or skill gap. The fix is study: more deliberate examination of the patterns you are trying to read, not changes to rules about sizing or timing.
Execution error. The analysis was sound but the action did not match the plan. The decision was made at the wrong moment, in the wrong instrument, or in the wrong direction relative to what the analysis specified. Execution errors are often mechanical — a misread, a hesitation, a click in the wrong direction. The fix is procedural: slower confirmation steps, an explicit pre-action checklist, removal of the friction points that cause mismatches.
Risk error. The plan was identified correctly and the action matched the plan, but the exposure taken was inconsistent with the sizing rules that govern the approach. Too large, too concentrated, held too long relative to the pre-defined limit. Risk errors are rarely about analysis. They are about adherence to a structural rule. The fix is rule-based: a fixed sizing formula applied before entry, not re-evaluated in the moment.
Psychology error. The plan was clear, the sizing was correct, and the action was taken — but the decision to enter, hold, or exit was driven by an emotional state rather than by the conditions that define the edge. Fear of missing a move, impatience during a wait, reluctance to close a losing position because it feels like admitting failure. Psychology errors are detectable by asking: would I have made this exact decision if my emotional state had been neutral? If the honest answer is no, the error is psychological. The fix is a pre-session state-check routine — a short, written assessment of readiness before any decisions are made.
Process error. The rules themselves were not followed, not because of emotional pressure, but because the process was unclear, incomplete, or inconsistently defined. If you cannot state your criteria precisely enough to test whether a decision violated them, that is a process gap — not a psychology gap. The fix is documentation and specification: write the rules precisely enough that a second person could follow them without asking you questions.

The Extra Bucket: Variance

This is the category that matters most for accurate diagnosis, and the one most often omitted. When a decision followed the plan correctly — analysis was applied consistently, execution matched the intention, sizing was within the rules, no emotional state distorted the choice — and it still lost, that outcome gets no error label. It is variance.

Labeling variance as an error is one of the most common and most damaging mistakes in self-review. It produces false diagnoses. A trader who assigns an error label to every loss will accumulate a distorted tally — one that suggests far more process failure than actually exists, and that directs effort toward fixing things that are not broken. It also trains a subtle form of outcome-bias into the review process itself: the label becomes a backward inference from the result rather than a forward judgment of the decision.

Variance is not a consolation prize. It is a precise classification: the plan was sound, it was followed, and the market moved against it. That is a different event than any of the five error categories, and it deserves to be recorded as such. Over a large enough sample, a well-designed process should produce variance losses at a rate consistent with its stated win expectancy. If variance losses are rare and error-labeled losses dominate, the process has real problems. If variance losses match expectation and error-labeled losses are infrequent, the process is functioning.

Counting Across Decisions: Finding the Dominant Type

A single tagged decision tells you almost nothing. The diagnostic value of this taxonomy emerges only when you count across a meaningful sample — a minimum of fifteen to twenty closed decisions, ideally more. The procedure is simple.

After each closed decision, assign exactly one label: analysis, execution, risk, psychology, process, or variance. Do not assign multiple labels to a single decision — choose the primary failure. Keep a running tally by category. After your sample is large enough, calculate the proportion of labeled errors (not variance) that fall into each category.

The category with the highest count is your dominant error type — and it should drive your entire practice focus until the count shifts. If analysis errors lead, your sessions should be spent studying and drilling recognition, not reviewing your sizing rules. If psychology errors lead, your sessions should begin with a state-check, not with chart study. If risk errors lead, your sizing rule needs to become automatic and pre-committed before you open a session, not deliberated during it.

This is why the fix differs by category. There is no general solution to "making mistakes." There are only specific solutions to specific categories of error. Counting is what reveals which solution applies.

A Hypothetical Illustration

Imagine a trader who reviews thirty closed decisions from a recent stretch. She labels each one and tallies the results: six analysis errors, four execution errors, two risk errors, nine psychology errors, two process errors, and seven variance outcomes. The dominant category is psychology — nearly a third of all labeled errors trace to emotional state distorting the decision. Studying chart patterns more intensively will not help her. The intervention that matches her dominant type is a pre-session readiness routine: a written check of her state before any decisions, and a hard rule that she will not act while that check is in a flagged condition. The other categories exist, but psychology is what to address first.

If she had simply reviewed the outcomes without categorizing, she might have focused on the analysis errors because those produce the most visible "wrong reads" — and missed the pattern that actually drives the majority of damage.

The Discipline: Tag First, Prescribe Second

The rule is sequential. You do not prescribe a fix until you have counted across a sample. You do not count until you have applied labels. You do not apply labels until you have separated the decision from the outcome. That sequence protects against the most common failure in self-review: jumping to a solution before confirming the diagnosis.

Keep the taxonomy small enough to apply consistently. Five categories plus variance is the right level of resolution. Finer distinctions — sub-types of psychology errors, sub-types of analysis errors — are useful once you have confirmed the dominant category, not before. Start coarse, then drill down only where the count points.

Where This Taxonomy Comes From

The distinction between execution failures and plan failures was formalized not in trading but in safety science. In his 1990 book Human Error (Cambridge University Press), safety scientist James Reason showed that errors in complex systems are not uniform: slips and lapses are failures of execution — the right plan carried out wrongly — while mistakes are failures of the plan itself. Because the remedy for a slip (a procedure check, a slower confirmation step) does nothing to address a mistake (a flawed mental model of the situation), mixing the two categories together reliably produces misdirected fixes. Reason's framework was taken up in aviation and medicine precisely because it gave practitioners a way to ask the right question first: which type of error occurred? The same sequencing logic applies to trading self-review — a trader who repeats chart study after an execution error, or tightens a checklist after an analysis error, is applying the wrong remedy not from laziness, but because the diagnostic step was skipped.

Simulator Exercise: Tag Five Decisions

After a Speed Run session in Abu Terminal with at least three closed decisions — five or more is better — open your decision journal. For each closed decision, write one sentence describing what actually happened in the decision, then assign exactly one label from the taxonomy: analysis, execution, risk, psychology, process, or variance.

Do not adjust the label based on the outcome. Apply it based on the quality of the decision process at the moment it was made. If the session produced no labeled errors — if every decision genuinely followed the plan and the variance rate matches your stated expectancy — that is a valid and useful result. It means the process held under simulator conditions.

After tagging all decisions, tally the labels. If one category appears more than once in a small sample, treat it as a preliminary signal worth watching across future sessions. If the same category leads consistently across three or more sessions, it becomes your next practice focus. Assign one category at a time. Trying to fix two or three simultaneously dilutes the correction and makes it harder to confirm whether anything changed.

Note on simulator limits: the simulator removes real financial stakes, which reduces the intensity of psychology errors in particular. Errors that appear infrequently in simulator sessions may appear more often when real stakes are involved. Treat simulator tagging as a baseline, not a ceiling.

Error Taxonomy: Classify the Mistake Before You Fix It