A green week can be badly executed, and a red week can be cleanly executed. The result tells you what happened to the account; a weekly process audit tells you whether you followed the decision process you meant to test. Keep those two scores separate or a lucky rule break will look like skill while a well-executed loss will look like failure.
This page has one narrow job. The trading strategy report card evaluates a strategy over a declared evidence window. This four-category audit evaluates one week of execution: plan adherence, setup selection, risk execution, and state protocol. It does not certify profitability, diagnose psychology, or predict next week's P&L.
Disclosure: Trader's Second Brain is our product. The rubric works in any journal that preserves the required fields. TSB's current plan, import, review, export, coverage, and verified-date facts render from the server-owned catalog; the weights and editorial method below do not update automatically with it.
Quick answer: freeze the week, account scope, trade unit, rules, category definitions, weights, and grade bands before looking at results. For every category, show eligible n, pass count, missing count, and formula. Calculate P&L separately. Open the underlying trades before choosing one corrective rule.
Why P/L Alone Misleads About Weekly Trading Quality
Outcome and decision quality are related, but they are not identical. A profitable trade can violate the plan; a planned trade can lose. Baron and Hershey's original outcome-bias experiments found that people evaluated otherwise identical decisions differently after learning their outcomes. A later preregistered replication and extension reproduced the central effect in the tested decision scenarios. That evidence supports separating process review from outcome review; it does not prove that a particular trading rubric predicts returns.
The weekly audit therefore asks observable questions. Was the trade inside the written plan at entry? Did the setup meet criteria declared before the outcome? Was actual risk inside the stored limit? Was the trader's predeclared state gate completed? The P&L column stays visible, but it earns no points in those four answers.
Freeze the Audit Contract Before Grading
A weekly grade is only comparable when its contract stays stable. Write this header before calculating anything:
- Period: exact start and end timestamps plus timezone.
- Scope: included accounts, strategy version, instruments, and closed-position rule.
- Unit: one logical closed trade, one position, or another declared unit; do not mix executions with grouped trades.
- Evidence cutoff: when the export or query was generated.
- Eligibility: which fields must exist for each category.
- Missing-data policy: exclude, mark unknown, or fail the audit; never silently convert missing to compliant.
- Rubric version: category definitions, weights, grade bands, and safety caps.
The starting population must reconcile. If the week contains 40 closed trades, every category should explain why its denominator is 40, 39, 38, or another number. “Thirty-one compliant trades” is incomplete until the reader knows whether that means 31 of 31, 31 of 38, or 31 of 40.
The Four Weekly Audit Categories
The categories deliberately describe different questions. They may use the same trade rows, but a pass in one category never fills a missing value in another.
| Category | Evidence boundary |
|---|---|
| Plan adherence | Question: did entry, stop, target, invalidation, and allowed context match the stored plan? Needs: plan version plus reviewed trade. Do not infer: that a winner followed the plan. |
| Setup selection | Question: did the trade meet the predeclared setup eligibility rule at decision time? Needs: setup label/version and required inputs. Do not infer: that a clean-looking chart was eligible. |
| Risk execution | Question: was actual size and loss exposure inside the comparable planned-risk unit? Needs: planned risk, actual risk, and compatible units. Do not infer: that a small realized loss means risk was controlled. |
| State protocol | Question: was the trader-authored readiness or pause rule completed before entry? Needs: contemporaneous self-report or checklist evidence. Do not infer: emotion, intent, or diagnosis from timing alone. |
Category 1: Plan Adherence
Define a binary pass that can be audited. One workable rule is: fully followed only when the reviewed trade met the stored entry, stop, target, invalidation, size, and context rules. “Partially followed” remains useful evidence, but it is not a full pass if the numerator is called fully followed.
Formula: fully followed rate = fully followed eligible trades ÷ reviewed eligible trades × 100. Show partial and broken counts next to it. If a trade has no review, keep it in the starting population and put it in the exclusion row. The plan-adherence guide covers a deeper versioned-rule method.
A weekly rate does not establish discipline as a personality trait. It describes a declared set of records under a declared rule. The useful next question is which exact rule breaks repeated, not whether the trader “is disciplined.”
Category 2: Setup Selection
Write the setup contract before the week or freeze an already-versioned playbook. Required context might include session, market condition, trigger, invalidation, or a specific confluence definition. Those fields are strategy-specific; a universal count of indicators or risk-reward threshold would be invented precision.
Formula: eligible-setup rate = trades meeting the frozen setup contract ÷ trades with enough setup evidence × 100. A missing setup label is missing evidence, not a C setup. Grade the decision-time record, not a chart reconstructed after the outcome.
The practical check is simple: could another reviewer apply the frozen rule to the same pre-entry evidence and reproduce the classification? The trading-rules examples show how to turn a vague setup idea into testable conditions.
Category 3: Risk Execution
Risk execution compares planned exposure with actual exposure. Use one compatible unit within the category: account percentage, currency risk after conversion, contracts with the same contract specification, or R relative to the stored initial risk. Do not average incomparable lots, shares, and contracts.
Formula: within-risk rate = eligible trades inside the frozen risk limit ÷ trades with comparable planned and actual risk × 100. Count a moved or removed stop according to the rule that existed at entry, even if the trade later won. Record unavailable planned risk or incompatible units as exclusions.
Firm limits can be added as a separate hard gate when a funded program participates in the decision. The personal audit should not pretend a provider rule is static: resolve the exact current program from the canonical catalog and treat official terms as controlling.
Category 4: State Protocol
This category should grade an observable protocol, not claim to measure a hidden psychological state. Examples include completing a pre-entry checklist, respecting a post-loss pause, recording a self-authored readiness value, or stopping after a declared limit. A rapid re-entry may be observable; “revenge” is motive and requires trader-authored context.
Formula: state-protocol pass rate = eligible trades with the declared gate completed ÷ trades with contemporaneous state evidence × 100. If the journal has no structured field for your chosen gate, use a consistent tag or companion field and disclose that limitation. Do not reconstruct calmness from P&L or from a clean chart.
Define Grades and Weights Before Seeing the Week
There is no evidence-based universal A–F table for every trader. Build a local rubric from the behavior required by the strategy and account. This example is an editorial template, not a population benchmark:
| Band | Example declared score range | Meaning inside this rubric |
|---|---|---|
| A | 90.0–100 | Process held with limited exceptions |
| B | 80.0–89.9 | Usable week with a specific repair target |
| C | 70.0–79.9 | Material drift; review affected records |
| D | 60.0–69.9 | Process was not reliably reproduced |
| F | Below 60.0 | Rubric-defined failure or hard-gate breach |
Equal weights are the cleanest starting point when no category deserves priority. A trader may instead predeclare 35% plan adherence, 25% setup selection, 25% risk execution, and 15% state protocol. A prop-program hard breach may cap the total regardless of the weighted average. State the cap first; do not invent it after a bad week.
Worked Example: A Week That Actually Reconciles
Constructed example—not a customer result or a TSB-wide benchmark. Period: one declared trading week in the account timezone. Unit: logical closed trade. Starting population: 40 closed trades. Evidence cutoff: the export time recorded in the audit. The strategy and rubric versions were frozen before calculation.
| Category | Population reconciliation |
|---|---|
| Plan adherence | 40 starting = 38 eligible + 2 unreviewed |
| Setup selection | 40 starting = 36 eligible + 4 missing setup evidence |
| Risk execution | 40 starting = 39 eligible + 1 incompatible risk unit |
| State protocol | 40 starting = 34 eligible + 6 missing state evidence |
| Category | Pass and calculation |
|---|---|
| Plan adherence | 31 fully followed; 31 ÷ 38 = 81.6% |
| Setup selection | 34 matched contract; 34 ÷ 36 = 94.4% |
| Risk execution | 37 inside plan; 37 ÷ 39 = 94.9% |
| State protocol | 30 gate completed; 30 ÷ 34 = 88.2% |
With the predeclared 35/25/25/15 weights, the composite is:
(81.6 × 0.35) + (94.4 × 0.25) + (94.9 × 0.25) + (88.2 × 0.15) = 89.1. Under the example bands, that is B. The visible weak category is plan adherence, even though the composite nearly reaches A.
The score is not “89.1% profitable” and not evidence that the next week will improve. It is a compact summary of four declared pass rates with different eligible denominators. Keep the category vector beside the letter so the composite cannot hide a weak process.
Hidden Deal-Breaker: The Self-Grading Honesty Trap
The audit fails when definitions move after the outcome. Common forms are upgrading a winning rule break, downgrading a losing planned trade, excluding the worst record as an “exception,” or changing weights until the letter feels right.
The defense is mechanical:
- freeze the rubric and strategy version before opening weekly totals;
- make every category resolve to pass, fail, or missing under a written rule;
- keep missing rows visible and link the score back to exact trade IDs;
- calculate P&L in a separate block;
- version any changed definition for the next period rather than rewriting history.
The record-level workflow behind that discipline is straightforward: complete the review while the evidence is available, then keep the saved answer attached to the trade instead of reconstructing it from memory.
How to Run the Audit With TSB
TSB has processed 600K+ imported trades cumulatively. That is product-scale context—not the denominator for your week. The example above is exactly 40 closed trades, and every personal result must keep its own period, filters, eligible n, exclusions, unit, and denominator.
In the current Journal review flow, each reviewed trade can store an A/B/C execution grade, plan adherence as Yes/Partially/No, a main mistake, and an optional lesson. The Journal review panel reports reviewed and pending counts, the fully-followed rate, and the most frequent stored mistake for the active slice. CSV/JSON export can carry the review fields into an external audit.
Those fields support this method, but TSB does not silently turn them into the editorial four-category A–F composite above. Setup eligibility, comparable planned risk, and any state protocol still need declared definitions and adequate evidence. Where a category is unavailable, show Not verified or missing coverage rather than filling it from another field.
Audit the week from exact reviewed trades
Filter one account and period, close the pending reviews, export the evidence if needed, and calculate each category from its real denominator.
Open Journal review →Track Grades Without Moving the Goalposts
Trend only like with like: the same rubric version, account scope, unit, and category definitions. Put a version break on the chart when the strategy, broker route, risk unit, or scoring contract changes. A higher grade under easier rules is not improvement.
Always chart coverage beside the score. A week graded from 38 of 40 trades is not equivalent to a week graded from 12 of 40. If missingness concentrates in losing, unplanned, or emotionally difficult trades, the missing-data pattern may be more important than the composite.
Review P&L next to—but not inside—the process grade. Over multiple comparable periods, you may explore whether higher process scores coexist with better outcomes, but describe association rather than causation and retain every week. Do not choose the time window after seeing the line.
Turn the Lowest Category Into One Test
- Open the failing records. Name the exact trade IDs and rule that failed.
- Check concentration. Inspect setup, session, account, preceding outcome, and available notes without inventing motive.
- Choose one reversible rule. Examples include a fixed size cap, a checklist requirement, or a documented pause gate.
- Define the later comparison. Set the next period, eligible population, pass metric, and minimum coverage before testing.
- Preserve the baseline. Never delete or re-grade the original week to make the intervention look effective.
The weekly trading review turns the audit result into one Current Focus rather than five simultaneous behavior changes.
When a Weekly Letter Grade Is the Wrong Output
- Too little evidence: show the raw counts and missingness instead of forcing a grade.
- No written plan: create the strategy and risk contract before grading adherence.
- Mixed units: split incompatible accounts, instruments, or risk measures.
- Long holding periods: use a longer audit window that matches the decision cycle.
- Changed rubric: start a new version; do not compare the new letter directly with the old one.
- Purely subjective state evidence: report the self-assessment as such and avoid a pseudo-objective percentage.
Methodology Note
- Protected identity: the URL, title, four-category weekly audit, A–F interface, process-versus-outcome distinction, honesty trap, worked calculation, and improvement loop remain.
- Distinct intent: this page grades weekly execution. The strategy report card grades the evidence and risk of a strategy version.
- Evidence boundary: outcome-bias research supports separating known outcomes from decision evaluation; it does not validate universal trading thresholds or guarantee later P&L.
- Removed claims: unsupported inflation percentages, fixed trade-count requirements, guaranteed convergence windows, causal predictions, universal trader-type weights, and automatic compounding claims are not retained.
- Product boundary: TSB's implemented review fields and summary counts are described exactly; the editorial four-category composite remains user-defined.
For sourcing, correction, and evidence rules, see our editorial methodology.
Final Verdict: A Grade Is Only Useful When It Opens to the Trades
Keep the weekly audit—but make the letter the index, not the evidence. A useful report card lets the trader open the period, denominators, exclusions, failed category, and exact records behind the score. It preserves a profitable rule break as a rule break and a planned loss as valid execution evidence.
The strongest output is not “I had a B week.” It is: “Under rubric v1.2, 31 of 38 reviewed trades fully followed the plan; the 40-trade starting population included two unreviewed records; plan adherence was the lowest category; next week tests one declared repair.” That statement is specific enough to verify, act on, and later compare without losing the truth.
