Capstone: The One-Page Policy

The eight lines, and the number each one needs

A policy is a page, and its discipline is that every line carries a number or a checkable condition. The first line is the objective: what the account is for, over what horizon, and what would make the exercise a success even in a year that lost money — because a trader who has not said what the account is for will size it for something else. The second is the edge: the setups written out, each with its base rate and the location condition that the advanced rung showed can double the expectancy of an identical pattern. A setup list without rates is a description of a hobby. The third line is the unit, and it is the one that turns a view into an exposure: a percentage of equity, chosen so that the predicted losing run is survivable — the streak arithmetic from the advanced rung, not the trader’s appetite. The fourth is the exposure cap: the maximum risk open at once, so that a correlated day cannot add up to a position nobody chose. The fifth is the drawdown rule with all three of its parts — the trigger as a percentage from the equity high, the action such as halving the unit, and the restore condition — and it is written at the high, because the trade-off between ruin risk and recovery time cannot be resolved in the losing month. The sixth is the daily limit, which is a rule about the person rather than the market and exists for the pre-tilt state. The last two lines are the ones most policies omit. The seventh is the pre-mortem: the failure story written in advance, naming the specific ways this trader loses money — the chase, the doubled unit after three losses, the position held for the accounting — each paired with the if-then response from the previous lesson, so the failure has a cue and an action rather than a resolution. The eighth is the review: how often the record is read, what gets tagged, and the audit that prices the process in R, plus the unlock condition — the evidence that permits a larger size, stated before the trader wants one, so the increase is a conclusion rather than a mood. The page, with its numbers — Objective: What the account is for, over what horizon · Edge: Setups written out, each with its base rate and location condition · Unit: 1% of equity, chosen so the predicted streak is survivable ← · Exposure cap: Maximum risk open at once, so one day cannot exceed the plan · Drawdown rule: Trigger + action + restore, written at the equity high ← · Daily limit: Three units — a rule about the person, not the market · Pre-mortem and if-then: The failure story and a cue-response for each · Review and unlock: Cadence, tags, the audit in R, and the evidence that permits a size increase Every row ends in something countable. That is the test the page has to pass before it is a policy: a stranger should be able to take it and trade the account in your absence, including the part where they stop.

How the page is tested, and how it fails

Three tests decide whether the page is finished. The first is execution by a stranger: hand it to somebody who knows the market and ask them to follow it for a week, and every question they have to ask is a missing line. The second is countability: read each line and check that the trigger, the action and the restore condition are all present, because a rule with a vague chapter is a preference and preferences lose to states. The third is the audit, which is the only test that measures the policy rather than describing it — split a period by whether the page was followed and price the difference in R, exactly as the playbook lesson does with a year of trades. A policy whose gap is twenty-six R a year is worth maintaining; one whose gap is near zero needs rewriting rather than defending. The failures are predictable, and they arrive in three shapes. The first is length: a page that has grown into a document does not get read, and a policy that is not read is a decoration. The second is unmeasurable wording, which is usually a prohibition that sounds strong and cannot be checked — “only take A+ setups” is the standard example, and the fix is a written definition with a rate, not a sharper adjective. The third is missing restore conditions, which is what turns a rule into a trap: a trader who halved the size at the drawdown trigger and never wrote down what restores it will improvise a restoration, and improvisation is what the page was supposed to remove. The honest ending is that the page is a hypothesis about the trader rather than a promise about the market. It says: given the setups in line two, the sizes in lines three and four, the limits in five and six, and the responses in seven, this trader will produce the expectancy in the record. Most of that claim is testable within a year, and the tests are the audit and the calibration score, not the return — because the return contains the market and the process, and only one of them is being managed. When the audit says the process is not where the money is going, the page is wrong and gets rewritten. When the audit says the process is worth 26R and the year produced it, nothing changes, including after a losing month, which is the outcome bias the beginner rung priced and the reason the review reads decisions rather than results. • A stranger should be able to trade the account from the page, including the part where they stop. • Every rule needs a detectable trigger, a concrete action and a restore condition. • Keep it to a page; an unread policy is a decoration. • Replace prohibitions with definitions and rates — “A+ setups” is not a rule. • The audit is the only test that measures the policy rather than describing it. • Rewrite when the audit says the process is not the problem, not when a month loses money. The page can also be a way of avoiding the work: writing a policy is more comfortable than running one, and a document that is revised every month is usually a document that is never followed. The signal that the page is real is that it does not change between reviews, including in the week it is most inconvenient.

The handover test: could someone else run this page?

A policy has one test that is more demanding than any of its individual numbers, and it is operational rather than numerical: hand the page to a competent person who has never traded your account and ask them to make the decisions the page describes. Everything the page leaves implicit becomes visible. If the per-trade risk is stated as a percentage but the stop distance is not defined — from the entry, from the close, from a structural level — the new reader will compute a different size than you do on the same setup. If the heat cap is stated but the definition of heat is not, they will compute it on entries rather than on current stops and get a different number. If the drawdown trigger is stated but the portfolio it is measured on is not — strategy drawdown, account drawdown, drawdown from the high-water mark — the same losing month will trigger different actions. The exercise is not a formality; it is the only way to find out whether the page is a set of rules or a set of headings that you fill in from memory each time. The second thing the handover test exposes is the boundary of the page — the decisions it deliberately does not make. A policy that covers risk and not selection is complete for its purpose and silent about the largest discretionary choice in the process: which setups to take. That silence is fine, but it has to be stated along with where the decision is governed — the playbook, with its own written setups — and what the tiebreaker is when the page and the playbook disagree. In practice the ordering is simple: the page’s limits are hard constraints, the playbook’s setups are the permitted actions inside them, and where a setup requires a larger size than the page allows, the page wins. Writing that hierarchy down prevents the most common form of self-deception in a personal risk process, which is treating a limit as a consideration rather than as a constraint. The third is the **review cadence**, and it belongs in the page rather than in a habit. A policy that is tested is tested on a date, not when a mood suggests it, and the date should be fixed at the start — quarterly, or after a stated number of trades — with a written list of what is examined: whether every limit was observable, whether any breach occurred and what the action was, whether the risk fraction is still producing a drawdown the account can hold, and whether any line needs a change and on what evidence. The page then becomes a document with a version history, and the version history is the audit trail that makes the whole subject’s argument concrete: the state that produced a change is recorded, so a change made at the bottom of a drawdown is visible as such. What the capstone produces, when it is built this way, is not a plan that a person carries in their head — it is a mechanism someone else could operate, and that is the strongest available test that it is a mechanism at all. • Hand the page to someone who has never traded the account and see what they cannot compute. • Define the terms that decide size: stop distance, heat basis, drawdown reference, high-water mark. • State the boundary: the page constrains risk, the playbook governs selection, and the page wins a conflict. • Fix the review date in advance, with a written list of what is examined. • Keep a version history, so a change made inside a drawdown is visible as one. The test the page has to pass is not “does it contain the eight lines”, it is “does it decide without me”. Every ambiguity a second reader hits is a place where the decision was being made by whoever was holding the page in the moment — which is the arrangement this entire curriculum has been arguing against.

Sources

• Expectancy, R multiples and the trading planVan Tharp, “Trade Your Way to Financial Freedom”
• Performing a project pre-mortemKlein (2007), Harvard Business Review
• Implementation intentions and goal achievementGollwitzer & Sheeran (2006), Advances in Experimental Social Psychology
• Decisions versus outcomesKahneman, “Thinking, Fast and Slow”