What mistake analysis means
Mistake analysis is a structured way to examine a past decision to understand what went wrong and why, with the goal of improving future decisions. In practice, it usually involves: (1) describing the situation, (2) identifying what was done, (3) stating the expected outcome at the time, and (4) comparing that expectation to what actually happened.
Because it is retrospective, the method cannot directly observe all hidden variables. It relies on the inputs you collect (for example, the reasoning you wrote down, the time you acted, and the conditions your system experienced). That dependence on inputs creates the first major limitation.
How it works in practice
A typical workflow treats the decision as a set of assumptions. You reconstruct the conditions as you understood them, then evaluate whether the outcome was inconsistent with those assumptions.
Two parts matter for accuracy:
-
Stable mechanics vs. variable conditions Some elements are relatively stable—such as whether you followed a rule, whether your reasoning was internally consistent, or whether you evaluated information before acting. Other elements vary—such as market liquidity, volatility, execution quality, spreads, and any procedural constraints in the environment where the trade was executed.
-
Assumptions behind any calculation If you compare “expected” vs. “actual,” you are implicitly assuming something about what you would have gotten under the same conditions. When that assumption is wrong or incomplete, the analysis can blame the wrong cause.
Even without any real-time data, the analysis can still be useful for understanding decision-making. The limitation is that it may not explain results well enough to predict or prevent similar failures.
Evidence and example: why outcomes can mislead
Consider a simplified scenario: a decision performed poorly. Mistake analysis might conclude the problem was “bad judgment at the time.” But the poor result can also come from conditions that were not part of the original reasoning, such as execution slippage or different transaction costs than you modeled.
This creates a common failure mode: attribution error. The analysis assigns a single cause that fits the outcome, even though multiple factors could have contributed. When you cannot separate causes (because the missing data is not recoverable), the “what to do next time” conclusion becomes uncertain.
A second failure mode is selection bias. If you analyze only the decisions that went wrong, you can overestimate how often the same “mistake” causes losses. If you analyze only the ones you can easily reconstruct, you may miss the full range of decision contexts.
Limitations and risks
1) Uncertainty about the true cause
Mistake analysis can clarify what happened, but it may not prove why it happened. Hidden factors can include execution differences, timing effects, or limitations in the information available to you at the moment. As a result, the method can produce plausible explanations that are not verifiable.
2) Historical results do not establish future relationships
Even if you find a pattern—such as “decisions made under X conditions performed worse”—the relationship is not guaranteed to persist. Market conditions, participant behavior, and costs can change. Therefore, a past link between a decision style and outcomes may not hold later.
3) Conditions and costs can change the meaning of the same decision
Two decisions that look identical in reasoning can lead to different outcomes if the surrounding environment differs. Transaction costs, execution quality, and operational rules can make the same “mistake” appear more or less harmful. If your analysis does not account for these differences, it may overstate conclusions.
4) Overfitting to one narrative
Mistake analysis can become an attempt to fit a single story to complex outcomes. This can lead to narrow lessons that do not generalize. When you cannot test competing explanations, you may mistake a convenient explanation for the correct one.
Verification and next questions
To verify a mistake analysis, treat its conclusions as hypotheses, not facts. A practical check is to ask whether the explanation remains consistent across similar cases, using the same assumptions about conditions and inputs.
If you want to be more independent and accurate, clarify these questions:
- What information was actually available at the time of the decision? - Which parts of the environment varied, and were those variations recorded?