Module 6 · Lesson 7 of 17 · Risk, process, and mindset · About 5 minutes
The Weekly Review: Process Before Outcome
On this page
Before you read: Journaling Your Reasoning, Not Just Your Results
What you will learn
- How a scheduled weekly review uses the plan, checklist and journal together.
- Why process versus outcome is the main question.
- Why small samples and changing several things at once mislead.
What does a weekly review look at?
Once a week, what questions are worth asking?
Completed fictional review WR-01, D5 16:00 CST, for SIM-P v1.0, invented sessions D1–D5. This is the same record set as Journaling Your Reasoning, Not Just Your Results, not real history. Each entered simulation uses 10 units, $1/point/unit, entry 100, fixed total round-trip cost $2. The review preserves the final J03 clerical correction; it does not substitute the original mistyped +$19.
| ID | Exit | Gross P&L | Costs | Final net | Eligibility evidence |
|---|---|---|---|---|---|
| J01 | 98 | −$20 | $2 | −$22 | Followed; true gate before entry |
| J02 | 103 | +$30 | $2 | +$28 | Broken; false gate before entry |
| J03 | 102 | +$20 | $2 | +$18 | Followed; true gate; accounting correction retained |
| J04 | 99 | −$10 | $2 | −$12 | Unknown; before-record unavailable |
| J05 | 98 | −$20 | $2 | −$22 | Followed; true gate before entry |
Filled arithmetic: gross −20+30+20−10−20=$0; total costs 5×$2=$10; net −$10; mean −$2 per entered simulation. Two positive net results/5=40% sample win rate. Average positive net=(28+18)/2=$23; average loss magnitude=(22+12+22)/3=$18.6667. Sample expectancy 0.4×23−0.6×(56/3)=−$2, matching the direct mean. These are descriptions of five authored outcomes, not estimated future expectancy or proof of an edge.
Other completed counts: 5 entered cases; 2 no-entry observations N01/N02; total 7 decision records. N01 is known failed gate/no entry; N02 is insufficient-input/no entry. They are not wins, losses or zero-P&L trades in the five-trade denominator. Under the common stipulated arithmetic each entry would have a $22 planned amount against the dummy $25 comparison cap; this reconstruction does not prove J04’s missing before-note contained that plan or certify its unknown eligibility or guarantee actual losses could never exceed a threshold.
Before-entry evidence is available for 4/5 entered cases=80%. Eligibility is confirmed followed for 3, confirmed broken for 1 and unknown for 1. Adherence among known cases is 3/4=75%; reporting “75% adherence overall” without the unknown case would conceal the gap. Across all five, 3 are confirmed followed, one broken and one unresolved. Risk-size evidence and eligibility evidence are separate questions.
Review diagram
Read the diagram step by step
- Authored fiction. Education only; no observed results or predictive guarantee.
- Five entered simulations from invented sessions D1–D5; two additional no-entry decisions. Keep known process status separate from gain/loss and from missing evidence.
- KNOWN cases ONLY · n=4
- UNKNOWN · OUTSIDE the known-case grid
- J04: 1 loss, net −$12. Before-record unavailable. Missing evidence is NOT automatically a confirmed breach or compliance.
- Complete totals · all five entered cases
- Gross $0 − costs $10 = net −$10. Followed subtotal −$26 + broken +$28 + unknown −$12 = −$10. Five trades + two no-entry observations = SEVEN decisions. N01: failed gate/no entry; N02: missing input/no entry. Neither is a zero-P&L trade.
- 4/5 entered cases have before-evidence.
- Adherence ONLY among KNOWN cases
- 3/4 KNOWN cases followed; J04 remains unknown.
- Counts are NOT causal evidence
- The compliant losses do not prove the process wrong; the breach winner does not make a false gate good. This tiny fictional split does not estimate a causal dollar effect or a future edge.
- SIM-P v1.1 · ONE workflow change
- After WR-01, add an explicit BEFORE-snapshot-ID completion checkpoint at the top of every manual decision record. v1.0 already required evidence. Eligibility, quantity cap, exits and cost assumptions stay unchanged; no market signal added.
- NEXT five opportunities · 0 observed
- All five boxes are prospective, NOT completed records or review cycles.
- Evidence-quality check, not profitability
- Next 5 consecutive fictional opportunities, including no-entry decisions; log snapshot ID and creation time. All 5 need a retrievable before-record OR a contemporaneous missing-input/no-entry record. Reject missing/unresolvable/after-decision IDs or entry without its required before-record. Do not erase failures or extend until passing.
- Counts are derived from the shown five-record ledger, not the old invented ten-trade grid. Unknown evidence stays separate. One prospective workflow test is specified, not falsely reported completed.
Illustrative diagram constructed for teaching. Not real market data.
Counts are derived from the shown five-record ledger, not the old invented ten-trade grid. Unknown evidence stays separate. One prospective workflow test is specified, not falsely reported completed.
Why process before outcome?
Was a winning trade that broke the plan a good trade?
Use the known-case grid and a separate unknown row; do not force missing evidence into a favorable category.
| Confirmed status | Net gain | Net loss | Net subtotal |
|---|---|---|---|
| Followed | J03: 1 case, +$18 | J01/J05: 2 cases, −$44 | −$26 |
| Broken | J02: 1 case, +$28 | 0 cases, $0 | +$28 |
| Unknown — OUTSIDE known-case grid | 0 cases | J04: 1 case, −$12 | −$12 |
Subtotals −26+28−12=−$10 reconcile with the complete ledger. The compliant losses do not prove the process wrong, and the confirmed breach winner does not make the breach good. Nor does this tiny split demonstrate that compliance has a causal dollar effect: selection, chance, arbitrary toy conditions and missing evidence prevent that conclusion. Review the documented breach and record gap without changing definitions after seeing profits.
How should changes be made?
If something seems wrong, should everything change at once?
Filled decision: retain v1.0’s market-observation eligibility, quantity cap, exit references and cost assumptions for the next controlled practice period; no performance improvement is inferred. Make exactly one workflow change, version SIM-P v1.1, effective after WR-01: add an explicit before-snapshot-ID completion checkpoint at the top of each manual decision record, entered before proceeding with an entered simulation. v1.0 already required the evidence; v1.1 adds a conspicuous completion checkpoint to enforce retrieval before proceeding, rather than changing eligibility or adding a market signal.
Reason: J04 cannot be audited because its before-record is unavailable. A result-dependent strategy rewrite would not repair that evidence gap. J02’s false gate remains a documented breach, not a reason to optimize a new threshold from its +$28 result.
Completed next-period test specification — not yet observed: follow the next five consecutive fictional practice opportunities, including no-entry decisions, with v1.1 unchanged. Log snapshot ID and creation time for each; do not select only wins. Current completed follow-up observations=0 of 5. Acceptance check: all five have a retrievable before-record or a contemporaneous missing-input/no-entry record. Reject the workflow implementation if any required ID is absent, unresolvable or created only after the decision, or an entry proceeds without its required before-record. Do not erase failures or extend the window until it passes.
This is an operational evidence-quality check, not a statistically valid test of profitability or a required real trading cadence. Five complete records would not establish causal psychological benefit or a trading edge. Additional historical records/cycles have not been invented. If the evidence still cannot be captured, pause new entered simulations and fix record handling rather than manufacture a completed review.
What this does not tell you about reviews
- What your own review should conclude.
- How many trades make a sample large enough for you.
- Which changes, if any, anyone should make.
- Nothing here suggests any trade.
Method in Practice: one week, one change
Fictional example. John, Harbor Index futures, the prices, times, costs and results are all invented for teaching. This shows how a written plan can be built and tested. It is not a recommendation, and nothing here says how often this idea works.
Try it on a real chart. Test this idea yourself on past sessions with bar replay. Keep your own count, including the days with no trade.
The idea. The lesson says to review process before outcome and to change one thing at a time. Here is John's review of a five-session week with his plan from Volume Profile Explained: POC, Value Area, HVN, LVN. The one change is to how he follows the plan, not to the plan's rules.
Three outcomes
-
Process right, result good: Monday, +$75 (Replay A)
-
Process wrong: Thursday. He entered after one close, not two, and lost $70
-
Process right, no trade: Tuesday (Replay C) and Friday (release day)
Show the full review
| Day | Plan followed? | What happened | Result after costs |
|---|---|---|---|
| Mon | Yes | Replay A | +$75 |
| Tue | Yes | Replay C: no setup | $0 |
| Wed | Yes | Replay B: wrong-if met | −$104 |
| Thu | No | Entered after one close above 101.20 (101.30), at 101.35. Next close 101.10, out at 101.05 | −$70 |
| Fri | Yes | Release at 09:30: plan not active until 10:30, no setup by 12:00 | $0 |
| Week | 4 of 5 | −$99 |
Process first: one rule break (Thursday). Without it, the week would be −$29, but that isn't the point. The point is that Thursday's trade wasn't John's method.
The one change (version 2.1, dated Friday): "Before every entry, write the times and prices of both closes on the checklist." Why: Thursday's break came from counting a close that hadn't happened yet. What would undo it: if it adds errors or slows entries past the next bar's open in the next 10 setups. Not changed: any entry, exit or size rule.
Test it yourself
- Score process first: for each session, did you follow the plan? Only then add up money.
- Make at most one change, write it down with the date, and say what evidence would undo it.
- Make the change about the process unless you have held-back sessions to re-test a rule change on.
Where it breaks
- One week is far too short to judge a rule.
- A good week can hide rule breaks, and a bad week can be all good process.
- Too many changes at once make it impossible to tell what helped.
Reflection question: In your last review, did you look at the rule breaks before the money?
Check yourself
- Reconcile net total two ways.
- Why are J02 and J04 not both in the confirmed broken row? Hint: process error.
- What exactly changes in v1.1, and what would reject it?
- Does 40% sample win rate, or five clean future records, validate an edge?
- What are the four confirmed process/outcome cells in the review grid, and where does a case with insufficient pre-event evidence belong? Hint: plan followed or not, gain or loss.
- Why isolate one documented change rather than change several things at once, and what would this fail to prove? Hint: knowing which change mattered.
Reflect on your own process
- How does your weekly review keep trades that broke the plan separate from trades that simply lost money?
- When you change your process, how do you record which single change was made and when it took effect?
- How does your review count records with missing evidence, rather than quietly leaving them out?
- In your last review, did you look at the rule breaks before the money?
Sources
This lesson’s numbers are invented teaching examples. Their arithmetic was checked; no outside source is needed.