FIND YOUR ONE LEAK

The stage’s payoff: from your own tallies, the single recurring error pattern that costs you the most — worth more than any new strategy — and the one rule that retires it. Plus the board: which setups get promoted, which stay on probation, and which get archived with their numbers showing. A boat that takes on water does not need a faster bilge pump, a bigger crew of bailers, or — the market’s favourite suggestion — a second boat. It needs the seam found: the one place the water actually enters, which is rarely the loudest splash and never found by admiring the hull’s good panels. And the finding has a method: you don’t re-plank the whole hull every week (that way the boat is perpetually in dry dock and nothing is ever tested); you mark the seams, sail, and see which mark is wet again. Trading’s leak hunt is the same discipline: the journal marks every error pattern, the tallies show which mark keeps getting wet, and ONE seam gets sealed — properly, with a rule instead of a resolution — before the next launch. The other seams are watched, logged, and left alone, because a boat under constant total repair teaches its owner nothing about which repair worked.

THE ONE LEAK · STAGE 9 · LESSON 6

THE TAKEAWAYS

• The scoring, done on the sample so the practice can copy it exactly: pattern words with (count, total R) — EARLY ENTRY (2, −1.30R); REVENGE ENTRY (1, −1.00R); OVERSIZE (1, −1.00R); DOOR BEND (1, +0.02R). Read the table like the method does: revenge is already handled by a standing guard (Stage 8’s circuit-breaker — its fix exists and is mechanical, so it needs vigilance, not the goal slot); oversize gets a lock today (size copied from the division line — a ten-second change at the ticket); the door bend is flagged for watching precisely BECAUSE it paid — paid bends are how doors quietly become suggestions. Early entry is the only pattern with a count above one and a four-figure trajectory if annualised by habit: it takes the goal slot, the leash, and the next review’s first line. Frequency × price, with guards credited — that’s the whole algorithm.
• Why the leak outranks a new strategy, in the sample’s own arithmetic: the best available ‘new strategy’ evidence in the sample is the goes-up debit’s +0.83R over two trades — a hypothesis. The leak’s −1.30R over twelve is a measurement: those two trades happened, at those prices, for one stated reason (the trigger existed; the waiting didn’t). Retiring the leak is worth more than the sample’s entire clean profit (+0.43R) three times over, and it requires no market insight whatsoever — only a rule, obeyed. This is the craft stage’s central economic claim, and it is modest on purpose: across a real log, the recurring-error line is usually the largest single recoverable sum a trader owns, and it is recoverable by paperwork.
• Deliberate practice is aimed at the leak, not at ‘the market’: between trades, the sample’s trader rehearses THE TRIGGER on historical SPY charts — Stage 3’s discipline turned into drills. Twenty old shelf days, chart paused at the test: call ‘trigger / no trigger’ before revealing the next day’s bar; score it in the journal like a trade. The drill targets the exact muscle that failed (waiting for the held close) on evidence that can’t cost money, and its score is a process metric the weekly review can watch while the live count is still small. Practice aimed generally (‘study more charts’) diffuses; practice aimed at the leak compounds. When the leak changes — Lesson 4 watched it migrate from entries to an exit bend in one week — the drills re-aim at the 30-trade review, not before: one target per cycle, or the aiming itself becomes whack-a-mole.
• The board, and its constitution: verdicts are read off CLEAN lines (a setup is never convicted of its owner’s errors, nor acquitted by their luck), at the 30-trade review, with tallies every ten trades in between (Stage 4’s rhythm, promoted). PROMOTE — a candidate whose clean line at its own 30 supports it joins the book, with its page’s stats box finally filled from its own log. PROBATION — terms in writing: what the clean line must show, by which review, or the page retires. RETIRE — a setup whose clean line is negative at a real sample is archived: the page kept, stamped RETIRED with the numbers that retired it and the date; the archive is the tuition receipt drawer, and an archived setup returns only as a new paper candidate with blank stats — never by quietly resuming. The sample’s board at twelve trades is PROVISIONAL (the graphic says so): candidate, leash, probation — three paperwork states, zero executions, because twelve trades haven’t earned the court a verdict yet. That restraint is the system working.

WORKED EXAMPLE — STEP BY STEP

1. Score the sample’s patterns (from Lesson 5’s ledger): early entry (2, −1.30R, −$184); revenge (1, −1.00R); oversize (1, −1.00R); door bend (1, +0.02R, memo: forgone +0.13R of door value). Now apply the guards test: revenge’s fix already exists as a standing machine (the circuit-breaker, installed in Lesson 3’s review and restated in Week 2’s goal); oversize’s fix is a lock, installable today. Neither needs the goal slot. Early entry has no machine yet and a count of two: it is the leak. Verdict written in the book: ‘THE LEAK (cycle 1): early entry — −1.30R across trades #5 and #12. Rule: no ticket before a held close; plan card read aloud first. Judged at trade 30.’
2. Watch the two instances share a fingerprint, because the rule must name it: both entries were taken BEFORE the confirming close on a shelf test — the trigger was written, the level was real, the waiting didn’t happen. (#5 in a range week, #12 in a breakdown week — the regime changed; the impatience didn’t. A pattern that survives a regime change is a property of the trader, which is exactly why it outranks any property of the market.) The rule therefore attaches to the trader’s behaviour at the ticket, not to a market condition: held close, read aloud, then the ticket — in that order, in all weathers.
3. Install the companions mechanically, so the goal slot stays singular: the ticket’s size field is COPIED from the plan card’s division line (never typed), which retires oversize as a class without spending a goal on it; every exit not taken at a written door is tagged DOOR BEND the same day and priced at its forgone door value as a memo — the tag that would have caught trade #9’s paid bend before it became a habit. Mechanical fixes are installed the day they’re identified; only the leak gets the weekly goal, the PlayBook leash, and the drills.
4. Run the board provisionally, the way the 30-review will run it for real: GOES-UP DEBIT — clean line +0.83R over 2 trades, no errors on file: CANDIDATE, collecting evidence; promotion requires its own thirty with the clean line intact — two wins are a poster, not a proof. STAYS-UP DEBIT — clean −0.16R over 2 clean trades, error line −2.30R: IN THE BOOK, ON A LEASH (the leak’s rule IS its leash — one fix serving two masters, the pattern and the page). SHELF CREDIT — clean −0.22R over 4: PROBATION, terms written: ‘clean line at or above flat by the 30-trade review, or the page retires to the archive with its numbers showing.’ Nothing promoted, nothing executed — the board at twelve trades is a docket, not a sentencing.
5. Set the practice that feeds the next cycle: twenty historical shelf days drilled against the trigger (call it before revealing the next bar; journal the score), and the leak’s count pinned as the first line of every weekly review until trade 30. If the count reaches trade 30 unchanged — zero new early entries across eighteen trades — the rule graduates into the page as permanent text and the NEXT pattern on the scored table takes the slot. If it grows, the rule was a resolution in disguise and gets rebuilt with a stronger trigger. Either way the review at 30 has a yes/no question to answer — which is what separates a system from a diary of intentions.

STAGE 9’S GATE — FOUR ARTEFACTS, NO P&L THRESHOLD

The plan’s gate for this stage is artefact-based, in the freeCodeCamp shape this course has used since Stage 4: produce, from YOUR OWN paper log — (1) FOUR WEEKS OF TAGGED JOURNAL: every entry carrying its setup, regime and error tags, written at the sittings, with the daily reviews filed (Lessons 1 and 3). (2) ONE COMPLETE PLAYBOOK PAGE for a setup from your own log (the PlayBook Page, Downloads): thesis, context, trigger, size, doors, three filed examples, a verdict — its statistics computed from your log or honestly blank (Lesson 2). (3) ONE WEEKLY REVIEW naming a recurring error with its price in R and the trade numbers that prove it (the Weekly Review sheet, Downloads) (Lessons 4–5). (4) THE MONTHLY BOOK PAGE: four weeks summed — R, dollars, percent of account — with the process account and the error account side by side (Lesson 4). No artefact asks what you made; a learner whose month is red and whose paperwork is complete passes this gate ahead of a green month with a diary. Hold those four pages and the craft is installed: what remains is the 30-trade cycle, run for as long as you trade. Stage 10 — going live, small, optionally — is open now, directly after this lesson in the path, and this gate is its prerequisite: its readiness audit scores exactly these artefacts. Meanwhile the ongoing layer’s rhythm is already yours: reviews completed, PlayBook entries updated, one leak at a time.

COMMON BEGINNER MISTAKES

• Hunting the loudest loss instead of the repeating pattern. The −$396 was one revenge trade with a guard already installed; the leak was two quiet early entries summing −1.30R — frequency × price, or the hunt convicts the wrong trade.
• Writing the fix as insight (‘be more patient’) instead of a rule with a trigger (‘no ticket before a held close; plan card read aloud first’). Eulogies don’t execute at 11am on a red day; rules do.
• Whack-a-mole: a new ‘leak of the week’ every review. No rule accumulates a count, nothing is ever judged, and the patterns rotate unwatched. One leak per cycle, judged at thirty trades.
• Deleting a retired setup’s page. The archive is the tuition receipt drawer — stamped RETIRED with its numbers, it is both evidence and warning. Deleted, it can be ‘discovered’ again next season at full price.

CHECK YOURSELF

DO IT — NAME YOUR LEAK, SET YOUR BOARD, ASSEMBLE THE GATE

From your own tagged log: (1) score your error patterns — (count, total R) per pattern word — credit the guards already standing, and name your ONE leak in the book’s format: ‘THE LEAK (cycle 1): ___ — ___R across trades #___, #___. Rule: ___. Judged at trade ___.’ (2) Write that rule into the PlayBook page it belongs to (Lesson 2’s page) and install the mechanical locks for the single-instance errors the same day. (3) Set your board: every setup in your log listed as CANDIDATE, IN THE BOOK (with its leash), PROBATION (terms written), or RETIRED (archived, numbers showing) — each read off its clean line, each labeled provisional until its 30. (4) Assemble the gate’s four artefacts (panel above) and check them against what you actually hold. Paper only. The cycle now runs itself: tallies at ten, verdicts at thirty, one leak at a time.