Consider a hypothetical trade that follows the written entry, sizing and exit rules but closes at a loss. Review compliance separately from the result. Following rules does not prove those rules have an edge, and one losing outcome does not by itself establish that the decision was an error.

Why a good decision can still lose

First reconstruct the information available before the order: instrument, costs, exposure, entry criteria and the adverse scenario. Then check actual volume, fills, protections and rule-consistent changes. Finally assess the method itself. Exact compliance can coexist with a flawed or unprofitable rule, so “followed the plan” is narrower than “a sound investment decision.”

Financial profit does not remove a documented rule breach, and loss alone does not establish negligence. Worse execution may involve a gap, liquidity or a technical problem; establish the sequence before attributing a cause. A single order can also reveal a wrong symbol, multiplier or ineffective protection. It is not automatically irrelevant until a large sample accumulates.

What "resulting" is — and why it is such a trap

In author-hosted interview notes, Annie Duke uses “resulting” for treating the outcome as the sole test of decision quality. Baron and Hershey’s 1988 abstract describes decision-evaluation scenarios involving medical choices or monetary gambles; knowing a favorable outcome influenced judgments. Neither establishes a Forex edge or makes trading statistically identical to poker. The limited practical point is to inspect earlier information as well as actual results.

Judging a method by one outcome can encourage unplanned changes. Review changes against the written rules and current evidence. A losing streak does not make a win or an edge “due”; assess risk limits and the method’s validity without treating past losses as a forecast of the next trade.

Both groups improve, overall rate falls

A fictional log divides decisions into simple and complex using a predefined, unchanged definition. Plan compliance rises in both groups. Yet the overall rate falls because the second review contains many more decisions from the group with lower compliance. Before calling it worse execution, check the sample mix.

Scroll sideways to see the full diagram

Sample composition reverses the overall resultIn review one, simple decisions comply in 90 of 100 cases and complex decisions in 1 of 10. In review two, simple decisions comply in 10 of 10 and complex decisions in 60 of 100. Both groups improve, but the total falls from 91 of 110 to 70 of 110.Review 1Review 2Simple: 90/100 = 90%Simple: 10/10 = 100%Complex: 1/10 = 10%Complex: 60/100 = 60%Overall: 91/110Overall: 70/110≈ 82.7%≈ 63.6%Complex cases were 10/110,then 100/110.Groups ↑; overall ↓ as mix changes.
Bars compare rates within each group. Counts show why the second overall result weights groups differently.

Keep numerators and denominators beside percentages. Do not turn this demonstration into proof that your method works: small, differing samples can differ for many reasons. Compare compliance within fixed groups first, then their shares of the total. Separately assess breaches of material risk conditions that an average may conceal.

Assumptions: A deliberately constructed arithmetic demonstration, not an observation of traders. Group definitions and compliance criteria are identical in both reviews; no decisions were removed or reassigned after assessing outcomes.

Four boxes: decision versus result

The four combinations below separate rule adherence from a positive or negative result. They organize review questions; they do not identify the cause of a profit or loss or prove that compliance establishes decision quality.

Rule adherence and result: separate review questions
Rule followed, profitDid fills and costs match the assumptions?
Rule followed, lossDid the event reveal a method, execution or risk issue?
Rule breached, profitWhat enabled the breach and how can repetition be limited?
Rule breached, lossWhat is documented and what needs correction?
Two assessmentsSeparate recording does not establish statistical independence or a guaranteed long-run result.

A high compliance percentage can conceal a material breach. In an own hypothetical review, 18 of 20 trades follow the checked rules: 90%. One of the other two may breach the whole account limit. Record critical events separately; the 90% average does not neutralize them. Nor does a losing streak by itself make an assumed edge due to appear next.

A larger sample does not make one trade meaningless

A larger sample can reduce uncertainty under suitable assumptions, but there is no universal trade count that separates edge from chance. Dependence between trades, payoff variation, costs, changing conditions and repeated strategy selection all affect the analysis. Keep plan-compliance assessment separate from evidence of positive expected value.

Choose routine review intervals for the activity, but investigate technical problems and limit breaches when discovered. One entry can identify an incorrect contract multiplier or commission counted twice. Recency bias concerns how recent observations may influence judgment; it does not require waiting until a monthly or quarterly review to act on a documented fault.

How to practise judging by process, not result

Keep the rule version, pre-entry information, sizing rationale and execution record. Mark unavailable evidence “unresolved” rather than defaulting to compliant. Decisions may affect exposure and costs; separate assessment does not imply independence of process and outcome. Both result and method assumptions remain relevant.

  • Review the decision record. State what was known, which criteria were met and how volume and protection were chosen. Intention alone does not establish correct execution.
  • Keep separate fields. Record adherence, execution differences and net financial result. Keeping a trading journal supplies a record; the fields are not independent statistical variables by definition.
  • Use routine and event reviews. Broader method evaluation needs sample size, period, costs and changing conditions. A technical or risk fault can require action immediately; the trading psychology section provides related topics.
  • Do not make the result a personal verdict. A planned loss is not an admission fee buying an edge. Loss acceptance includes checking facts and existing exposure without a recovery obligation.

In an own hypothetical model, gains of 0.5R occur with probability 0.55 and losses of 1R with probability 0.45. Expected result is 0.55 × 0.5R − 0.45 × 1R = −0.175R before additional costs. R is one fixed planned-risk unit, not a guarantee of the actual loss. Compliance does not repair those unfavorable assumed parameters. More observations can reduce uncertainty about a mean under suitable assumptions, as NIST’s interval formula illustrates; that is different from reducing the variance of an individual trade or proving a durable edge.

What to do tonight

Before a joint process review, check how reviewers interpret the rule. In an own hypothetical set of eight cards, two reviewers agree on six: 6/8 = 75%. Card 03 differs over an exception; card 06 may use different rule versions. Agreement is not trade compliance or profitability: both reviewers may agree that a rule was breached. Without an independent reference answer, majority agreement does not prove correctness.

Retain both initial assessments and their evidence before discussion. Clarify the exception or version; record limitations in the old review and improve the rule prospectively, without rewriting it to make all past cards compliant. In a historical exercise you may temporarily hide the outcome field, then use fills and net results to test assumptions. Do not hide information needed for live exposure management. A growth mindset is a learning topic, not proof that a method survives its next losing streak.