Trading plan and setup playbook — rules, risk and review

A trading plan sets the boundaries for your work: which markets you cover, when you can manage positions, how much risk you may take and when to stop. A playbook describes the individual setups permitted within those boundaries. This lesson brings the two documents together through a nine-section plan, a thirteen-field setup card and worked examples. It also shows how to record decisions before the outcome is known, distinguish a rule from a research hypothesis and review results without rewriting the past. Completing the documents gives you something you can check and test; it does not establish that a strategy is profitable.

Illustrative plan and playbook artwork; the written rules and synthetic-data qualifications below remain authoritative.

1. Trading plan and playbook: two different documents

These documents are easy to confuse. Some educational material uses “trading plan” for a mixture of mission statements (“I will be a disciplined trader”), risk guidelines (“No more than 2% per trade”), strategy descriptions (“I will trade H4 breakouts”), and behavioral checklists. A document that becomes a home for everything can become difficult to use for any one task.

The distinction that makes them manageable is:

The practical difference: an example

In the trading plan, you write: “Maximum planned risk per trade: 1% of capital. Daily loss limit: 2%, equivalent to two full A+ losses under the period-opening-capital convention. Once reached, the session ends until the next permitted day.” That rule does not change because the instrument is EUR/USD rather than GBP/JPY, or because the strategy uses breakouts rather than mean reversion. It governs capital allocation.

In the playbook, you write: “EUR/USD London-open breakout: active when D1 ATR(14) is 60–120 pips, the pre-London range formed by the four M15 candles from 07:00–08:00 Europe/London is 12–35 pips wide, and there is no relevant high-impact release within the card's ±2-hour event window. Entry: price breaks the range by at least eight pips, with an M5 candle closing above the upper boundary for a long or below the lower boundary for a short. SL: five pips beyond the opposite boundary. TP1: half the position at 1R. TP2: the other half at 2.5R. A+ planned risk: 1%.” This is an operating instruction that can be tested. The full thirteen-field card is in section 4, and the same parameters must appear in the card, journal, and test. The time above corresponds to 08:00–09:00 Europe/Warsaw range; keep the named time zone rather than replacing it with an ambiguous fixed offset.

2. A+/B/C classification: mandatory conditions and quality criteria

Each setup in the playbook has its own outcome profile. A class is not a beauty score for a chart; it is shorthand for a predefined sizing decision. In this example, A+ means every mandatory condition passes and all quality criteria are comfortably within their accepted ranges. B means the same mandatory conditions pass, with exactly one quality criterion in a specifically defined borderline band, recorded before the outcome is known. C means a mandatory condition fails or more than one quality criterion is borderline. There is no entry in class C: record the rejection and its reason.

The classification has a practical purpose: preventing you from increasing size simply because you have not placed a trade for three days. Without predefined distinctions, everything on the screen can look equally attractive, leaving FOMO after inactivity or anxiety after two losses to decide the position size. The plan blocks that improvisation. The playbook reduces the number of decisions you need to invent while looking at a moving chart.

Layer 1: mandatory conditions are a gate, not points

Every mandatory condition must be YES. Otherwise there is no trade, whatever class you hoped to assign. Missing information counts as NO.

• Plan limits: daily, weekly, and monthly restrictions allow another entry; aggregate planned risk after entry remains within its limit; no prohibited correlation conflict exists with an open position.
• Calendar: no relevant high-impact release falls inside the event window specified on the setup card.
• Session and entry window: both match the card.
• Order ticket and record: SL and TP are calculated before submission, volume comes from the sizing calculation, and the pre-trade journal record is prepared before the outcome.

Layer 2: quality criteria determine the class

Illustrative thresholds, not market standards. The numbers below are assumptions consistent with the plan's risk section, not recommended settings for every reader. If sizing changes, update the plan, setup card, and copyable template together.

“Borderline” must mean a number on the card, not a feeling. The convention in this example is a band within 5% of the threshold, but on its permitted side. It is one-sided: a value beyond a disqualifying threshold is not “almost satisfied.” Field 8 of the London setup card lists the actual bands; the underlying definition belongs to the setup owner and the strategy-building process in article 15.3. Without a written band, “borderline” can mean whatever supports the entry you want. Classification is also subordinate to the mandatory gate and the card's invalidation rules. A disqualified setup does not become tradable as B, or at any other reduced size.

There is no “C at the absolute minimum size.” A smaller position does not override a failed mandatory condition: it remains a trade your playbook does not authorize. Count rejected candidates separately. If a quarter contains no rejections, investigate whether the conditions filter anything in practice; zero rejections is a diagnostic question, not proof by itself that the rules are defective.

Where the numbers come from: testing and the journal

A+/B/C is intended to be quantitative, not aesthetic. “This looks like A+” is not a classification method. Its evidence develops through:

1. The original backtest: an initial outcome distribution and proposed criteria distinguishing A+ from B. Article 15.4 covers testing methods, sample separation, and pitfalls.
2. Forward observation: on demo, test whether the rules could be executed in real time without seeing later candles. Demo does not reproduce live liquidity, rejected orders, requotes, partial fills, or slippage faithfully. Evidence about actual execution requires appropriately controlled live records if and when live trading is independently justified; moving to a small funded account is not an automatic course requirement.
3. A live-trading journal, where available: reassess the classification using actual costs and execution. A setup that appeared A+ in a backtest may behave differently after slippage, costs, and asymmetric fills.

If an A+ label has no trade count, mean R outcome, cost record, or drawdown information behind it, it is a classification hypothesis rather than demonstrated superior quality. Otherwise, the label can become a sticker on a picture, with everything subjectively looking good at entry.

3. A trading-plan template with nine required sections

Keep the working plan short enough to read before a session: approximately one to three A4 or letter-size pages is a useful target for this example. Page count is a usability guide, not a reason to omit a required rule or shrink the font. Give each of the following nine sections one to three sentences or a list of roughly three to seven points. A missing section leaves an unresolved decision that may otherwise be made under pressure.

Section 1: process objective and outcome scenario

The first objective is about the process, not a percentage return. It concerns actions you can check every week. For example: “During the first six to twelve months: no breaches of daily or weekly limits; every trade has a journal entry prepared before the outcome is known; no entries outside the playbook.” Review these three commitments every weekend, regardless of what the market did.

An outcome target, if used later, should be written as a scenario assumption, not a forecast or obligation to make the market pay. Its arithmetic must be coherent. Nine months at +2.2% and three at −2% give 1.022⁹ × 0.98³ − 1, approximately +14% over the year, using compounding rather than adding monthly percentages. At 1% planned risk and around 70 trades, a simple linear approximation associates 14% with roughly +0.20R per trade after costs. A +30% linear scenario at the same size and count implies about +0.43R. These are consistency checks, not exact conversions between a monthly compounded path and a mean trade outcome; the return path and sizing convention matter. Evidence for the required outcome distribution must come before the target is treated as plausible.

Section 2: mandate—the permitted scope and strategies

The mandate states what you are allowed to trade at all. Without it, every new instrument or internet setup can be declared “within the plan” because nothing excludes it. An example is:

• Market and instruments: the broker's specified FX or CFD contracts on EUR/USD, GBP/USD, and USD/JPY only. Other pairs are outside the mandate. Product classification and availability depend on the account and jurisdiction.
• Horizon: intraday, maximum holding time four hours, no weekend positions.
• Working method: manual entries in the morning window defined on the card, with any permitted next-session pending orders prepared in the evening; no automated strategies or external signals. A prepared pending order must not activate outside the permitted entry window.
• Approved strategies and versions: “Playbook v2.1, four setup cards, including LON_Breakout_EURUSD_M5 v0.1. A setup without an ID and versioned card is outside the mandate, however attractive its chart looks.”
• Restrictions: no additional deposits to offset a drawdown; no change of execution provider during the quarter without a documented plan review.

The mandate is easy to omit and costly to reconstruct. Adding a pair or a style “just to try it” without changing the mandate can leave a quarter's journal mixing two different processes in one outcome column. A clearly labeled research backlog keeps experiments separate from approved trading.

Section 3: hard risk limits

Use one set of numbers in the plan, pre-entry risk filter, and section 10 template. Illustrative thresholds, not market standards. If the values change after review, change them in every location together.

• Maximum planned risk per A+ trade: 1% of the defined capital base.
• Maximum planned risk per B trade: 0.5%.
• Maximum aggregate planned risk of open positions: 3%.
• Daily realized-loss limit: 2%, equivalent to two full A+ losses under the stated fixed period-opening base; reaching it ends the session until the next permitted day.
• Weekly realized-loss limit: 4%; reaching it pauses trading until the next Monday, regardless of the macro calendar.
• Monthly realized result at −5%: halve the normal sizing for the rest of the month. At −8%: stop for the rest of the month and review the setups using article 15.10.

Denominator and measurement time. Period limits use capital at the start of that period: day-opening capital for the daily limit, week-opening capital for the weekly limit, and month-opening capital for the monthly limit. The measure here is realized performance from closed trades; open positions are addressed separately through aggregate planned exposure and the plan's incident rules. Reaching the threshold triggers the rule—an exact hit counts, not only an overshoot. The fixed-base convention means two full 1% A+ losses equal 2%, rather than the 1.99% obtained by resizing the second trade to reduced equity. Use compounding when examining a compounded equity path over many trades, and say explicitly when you change conventions. The card must also identify the base used to size an individual trade so the amounts can be reproduced.

What “1%” does not mean. It is planned risk under the specified stop-price and cost assumptions, not a guaranteed maximum loss. A gap, slippage, or costs omitted from the initial risk unit can produce worse than −1R. Similarly, “3% aggregate open risk” sums planned risks; it does not cap stress losses at 3%. A stop is an order subject to execution conditions, not a guaranteed price. State that in the plan so a neat percentage is not mistaken for a promise.

The limits must fit your circumstances rather than a handbook. Consider an illustrative PLN 5,000 account using 0.5% risk with a PLN 100,000 account using 1%; account size alone does not determine suitable risk. A consistency check: if 100 trades average −0.15R in a weak year, 1% nominal risk produces about −15% on a linear calculation. Repeated −0.15% returns would compound to roughly −14%, although the actual result depends on the distribution. “1% × 100 trades” by itself is not a return calculation: it totals 100 units of planned risk, not the percentage lost. If the downside implied by the scenario is unacceptable, revise the risk assumptions before relying on them.

Section 4: available time and trading hours

Establish how much time you have before deciding when to trade. A plan requiring unavailable hours will fail in its first week. For example: “Ninety minutes in the morning from 08:00–09:30 Europe/London, thirty minutes in the evening for preparation and permitted next-session pending orders, and ninety minutes at the weekend for review. Outside the morning window, use alerts rather than continuous chart observation.” This preserves the original 09:00–10:30 Europe/Warsaw example while anchoring a London setup to London time.

• Active entries: 08:00–09:30 Europe/London, matching the card. The London session continues much longer and later overlaps New York, but that does not extend this plan's entry permission.
• No new entries from 09:30 until the next permitted 08:00 window. With a four-hour maximum holding time, an entry at the end of the window must be closed by 13:30 London time at the latest. No evening or overnight exposure is intended; any evening-prepared order must be restricted to the next allowed window.
• Flat during the relevant event-exclusion window: in this article's example, ±2 hours around high-impact employment, CPI, FOMC, ECB, BoE, or BoJ announcements, as applicable to the card. The plan, card, and risk filter must use the same window.
• Flat during the defined rollover window. The original Polish example uses 23:00–00:30 for a 17:00 New York rollover in most weeks. For international use, record the broker's actual rollover convention and America/New_York as the event time zone; US and European daylight-saving transition dates can temporarily change the local conversion by an hour. This intraday mandate should already have closed its exposure much earlier.

Every time needs a time zone. “09:00” without one is not a rule another person can reproduce. Reproducibility by someone else is the test of an unambiguous instruction.

Section 5: the five-step entry procedure

1. Identify the setup in the playbook: which card and which version?
2. Check mandatory conditions: the card's regime, calendar, correlated open positions, and section 3 limits.
3. Calculate position size with the specified model, not by eye.
4. Prepare the order ticket with the intended SL and TP before confirming the order.
5. Prepare the journal record: setup, version, class, parameters, and planned R.

Complete the five checks before clicking Buy or Sell, then add the actual acknowledgment and fill details after submission. This clarifies the distinction between preparing the ticket and sending it: a journal entry cannot be made before a fill that has not happened, but its decision fields must already exist. Skipping these checks is precisely the kind of impulsive shortcut that pressure encourages.

Section 6: management of an open position

• Break-even moves, partial exits, and trailing rules must come from the specific setup's excursion and outcome analysis, not an assumed universal standard. Article 15.9 covers MAE and MFE. An accepted change becomes a new card version.
• Under this example's rule, never widen the SL after entry; a permitted stop adjustment can only reduce the originally planned price risk.
• A TP change during the trade is allowed only if its trigger was on the card before the position opened. A reason invented afterward is rationalization, not a predefined procedure.

Section 7: STOP conditions—when trading pauses or ends

• After two consecutive losses of either class: pause for 60 minutes and check for tilt.
• If the day's realized result reaches −2%: end the session, regardless of trade count. This stricter stop overrides a shorter pause.
• After three consecutive B losses, approximately −1.5% on the fixed base: end the session even though the daily percentage limit has not been reached.
• After reaching the weekly −4% limit: pause until the next Monday.
• A provider-side control where available: check whether the broker or prop account offers a configurable server-side loss limit, and understand exactly what it measures and does. Not every provider offers one, and a prop-rule threshold is not necessarily an automatic protection service. If unavailable, use the documented procedural alternative: handle open positions under the plan, cancel unwanted pending orders, and only then log out and record the stop. Closing the terminal does not close server-held positions or orders.
• A warning relative to your own normal state: unusually elevated arousal, a noticeably shortened interval between decisions, an impulsive reaction to a fill, or an immediate urge to recover a loss triggers at least a 30-minute pause and a reread of the plan in this example. Do not impose a universal heart-rate cutoff; caffeine, stairs, medication, and many other factors affect it. These are process cues, not medical diagnoses.

A rule kept only in your head can fail under pressure even when you know it^[7]. Write down the kill switch, define an observable trigger, and where possible use a control that does not depend entirely on the next discretionary click. Resuming depends on the documented process and elapsed stop period, not on winning the money back. Complete the required review and respect the day- or week-level pause. Article 15.8 covers what to do if a limit has already been breached.

Section 8: behavioral restrictions and no-trade rules

• No trading after less than six hours of sleep in this example. This is a personal process threshold, not a universal medical standard.
• No trading under the influence of alcohol.
• An external-stressor rule: a predefined disruption—family illness, a serious operational problem, a dispute, or a sleepless night—means the specified half-size regime or 24 hours flat. Decide which branch applies in the written rule before the session, not while looking at an attractive chart.
• Use a focus arrangement that mutes social notifications during trading while preserving essential broker and operational alerts.

A list of prohibitions is not enough. Each no-trade rule needs an observable trigger, a stated basis, an action, any permitted exception, a record location, and an owner authorized to revise it. Otherwise, within two weeks it may be unclear whether it still applies. The same format can audit every plan rule:

An exception not written into the applicable rules does not exist. “I will make an exception today” during the session is a breach, not a new authorized exception. Add any proposed exception during review, with its rationale, date, and plan version.

Section 9: records, review, and the change rule

Records. Every trade and every rejected candidate goes into the journal, with the decision recorded before its subsequent outcome is known. Article 15.9 defines the minimum fields, reconciliation to provider statements, and metrics. The plan needs one clear requirement: a journal without the setup ID and card version cannot reliably answer version-specific questions about the playbook.

Review. Use separate rhythms for separate decisions:

• Daily: record each trade and rejection.
• Weekly, around 30 minutes: identify rule breaches. Ask “Did I execute the plan?” rather than only “Did I make money?”
• Monthly: calculate setup and cost metrics using the definitions in article 15.9.
• Quarterly: review each card for retention, observation, downgrade, upgrade, or suspension, using the evidence criteria in article 15.10.
• Annually: review the whole plan—mandate, limits, available time, and stop conditions.

Change rule. Revise the plan or playbook only at a scheduled review or after an event on the predefined trigger list: changing provider, a capital change greater than half the reference amount, a material change in available time, or an operational incident. Each accepted change receives a new version and a concise rationale, while the previous version is archived. Do not rewrite a rule merely in response to one loss or a losing streak. Losing runs are a possible state of the system, not automatic evidence that a rule needs replacing. Pausing because of a genuine safety or operational problem remains allowed while the evidence is investigated.

4. The setup card: thirteen fields that make it usable

A setup can occupy one working page or one card in Notion, Obsidian, or a spreadsheet. Keep the summary readable before entry while retaining the necessary specification and evidence in linked records. Too little detail creates ambiguity; too much unstructured text becomes difficult to consult. Each of these thirteen fields answers a different question. The historical Turtle rules are a familiar example of instructions written so that someone other than their author could execute them^[5].

The name belongs in field 1 and should identify when, how, and what: for example, “London Open Breakout—EUR/USD.” That makes the catalog sortable and helps distinguish cards that could otherwise be confused.

A completed card: London Open Breakout—EUR/USD

Synthetic teaching example, not historical results. The following synthetic card demonstrates the recording format, not a profitable strategy or someone's account. Its pre-London window, ATR regime, stops, and targets are strategy specifications. Article 15.3 covers how specifications are developed and challenged. The governing rule is that the card, journal, and test must use the same parameters. Three different ATR thresholds in three places make the setup irreproducible, however precise each individual sentence sounds.

1. ID and version. LON_Breakout_EURUSD_M5 v0.1, effective from [YYYY-MM-DD]. Copy the identifier and version from the setup owner's specification in the strategy-building record. Version 0.1 is frozen before testing, so there is no earlier version to archive yet. Version 0.2 requires a recorded rule change, such as narrowing the entry window. Trades under that new specification must not silently enter the same sample as earlier trades.

2. Hypothesis. During the hour before London opens, EUR/USD may form a relatively narrow range. The research hypothesis is that the arrival of London activity can resolve direction and sometimes produce continuation during the following minutes. It is falsifiable: if breakouts under the defined normal-volatility conditions do not show positive net expectancy with your execution arrangements, the proposed mechanism has not established a usable edge for you. Archive or reject it under the research rules rather than endlessly adjusting it until a favorable history appears.

3. Context. D1 ATR(14): 60–120 pips, excluding the example's very quiet and shock regimes. Session: the first 90 minutes after the London open, 08:00–09:30 Europe/London, corresponding to 09:00–10:30 Europe/Warsaw. Instrument: EUR/USD. Article 15.6 covers regime indicators and validation of their thresholds. The card states the one threshold copied from the setup owner's specification and used in the test.

4. Mandatory conditions. All must be YES; missing data means NO:

• D1 ATR(14) between 60 and 120 pips.
• The pre-London range consists of the four M15 candles from 07:00–08:00 Europe/London, or 08:00–09:00 Europe/Warsaw. Its high–low width is 12–35 pips.
• The confirming M5 candle's tick volume exceeds 1.2 times the mean of the preceding 20 completed M5 candles, measured on the execution provider's feed.
• No relevant high-impact publication lies within ±2 hours of the proposed entry, for either currency. Checking only the London opening time would miss an event that enters the exclusion window before a later entry. Recheck at the actual decision time, consistently with 15.3 and 15.6. The plan's additional flat-before-event rule also remains binding.
• No existing position has the prohibited same-direction USD exposure through a pair such as GBP/USD, USD/CHF, or USD/JPY. Determine the exposure from direction, not the pair name alone.
• Plan limits allow an entry and the order ticket is complete.

5. Trigger. Price extends at least eight pips beyond the pre-London range and an M5 candle closes outside the corresponding boundary, above for a long or below for a short. Its tick volume must exceed 1.2 times the prior-20-candle mean. The specified entry is a market order after the confirmation candle closes: a long buys at the available ask and a short sells at the available bid. The actual fill is not guaranteed to equal the chart close. There is one entry method here. A limit order waiting for a pullback to the boundary would be a different rule, version, and sample.

6. Invalidation. The setup is unavailable for the day if: (a) no qualifying M5 close has occurred by 09:30 Europe/London; (b) price returns inside the range before the prospective confirmation candle closes under the written invalidation rule; or (c) an unscheduled relevant announcement is identified within the excluded event context. Record invalidation as a rejection; it is data, not an absent trade. Invalidation overrides quality classification. An invalidated candidate is not downgraded to B and traded at half size.

7. Entry and exit. Place the server-side SL five pips beyond the opposite range boundary. TP1 closes 50% at 1R from entry. After TP1 fills, move the remaining stop to the entry price: a subsequent exit there gives +0.5R gross on the original position, not zero, before costs. TP2 closes the remaining 50% at 2.5R. Only after TP1 has filled, apply the specified M15 SMA-20 trailing rule: after each completed M15 candle, update the remaining stop toward the current SMA-20 level only when that reduces risk, never below entry for a long or above entry for a short, and subject to valid broker stop placement. The remainder exits at whichever happens first: TP2, the accepted trailing stop, or the time exit. If a historical bar touches both stop and target and the sequence is unknown, use the same pessimistic stop-first convention as the test; that is a simulation assumption, not a claim about live event order. The rules impose a gross ceiling of 0.5 × 1R + 0.5 × 2.5R = 1.75R under the stated target-fill assumptions. Trailing can reduce the achieved result; it does not raise that planned ceiling. The maximum holding time is four hours, after which the example closes the remainder and records “no follow-through.” The controlling specification is the setup owner's card in article 15.3.

8. Sizing model and quality criteria. A+ uses the plan's 1% risk, B uses 0.5%, and C has no entry. A calculator derives volume from the stop distance and the defined risk budget. The card quotes the plan rather than establishing another sizing rule. Three named quality measurements determine the class:

• Structure: range width, accepted at 12–35 pips. Borderline bands: 12.0–12.6 or 33.3–35.0 pips, using the card's stated rounded upper-band boundary.
• Momentum: confirmation-candle tick volume relative to the preceding 20 M5 candles. Required ratio: strictly greater than 1.2. The borderline band is above 1.20 through 1.26; exactly 1.20 fails the strict gate.
• Context: D1 ATR(14), accepted at 60–120 pips. Borderline bands: 60–63 or 114–120 pips.

The 5% convention applies only on the accepted side of the threshold. A range of 11.4 pips, volume ratio of 1.14, or ATR of 57 or 126 is outside the gate, not “nearly acceptable.” A symmetric ±5% tolerance would quietly broaden the strategy. These definitions belong to the setup specification, not to a last-minute interpretation in this article. Three accepted measurements with none borderline produce the illustrative A+ classification; exactly one borderline produces B; a failed threshold or more than one borderline means rejection. Record class before entry. The same measurements appear in field 4 for a reason: the gate asks whether a value is allowed, while the quality criterion asks how close an allowed value is to the boundary. Two questions about one measurement are not two independent thresholds.

9. Costs. The example cost model is two pips of spread plus half a pip of slippage: 2.5 pips of friction. When that friction is already reflected in the entry and exit prices used for the outcome, do not subtract it again. Relative to the 35-pip initial stop in field 11, 2.5/35 is approximately 0.07R. Across the illustrative 25–48-pip stop-distance range, it is approximately 0.05–0.10R; the exact distance also depends on the entry extension, so that illustration is not an additional unlisted stop cap. Commission and any financing are separate net-outcome deductions outside the card's price-distance R0. A synthetic USD 6 round-trip commission per standard lot corresponds to 0.6 pips at USD 10 per pip, or about 0.02R for a 35-pip stop. R0 here is entry-to-initial-SL price risk before commission. Use that same convention in the journal so “net” means the same thing in both places.

10. Prohibitions. No entry after 09:30 Europe/London. No second attempt in this setup on the same day, including after the first attempt was invalidated. No widening the SL after entry. No size increase because of a winning streak. No “smaller-lot” entry when a mandatory condition is NO. Excluded days: the Employment Situation release day and the FOMC decision day, identified from the publishers' current-year calendars rather than a memorized weekday rule.

11. A valid-format example—synthetic weekday prices during the London session. D1 ATR(14) is 78 pips: accepted and not borderline. The 07:00–08:00 Europe/London range has high 1.08420 and low 1.08210, a width of 21 pips: accepted and not borderline. No relevant high-impact event lies inside the exclusion window. At 08:20 London time, the 08:15–08:20 M5 candle closes at the example's 1.08510 level, nine pips above the range high, exceeding the eight-pip trigger. Its tick volume is 1.6 times the prior-20-candle mean: accepted and not borderline. The synthetic ask fill is 1.08510. SL is 1.08160, five pips below the range low. The 35-pip distance defines R0, with the illustrative A+ budget at 1% of the stated capital base. TP1 is 1.08860; TP2 is 1.09385. The four-hour deadline is 12:20 Europe/London, corresponding to 13:20 Europe/Warsaw. Before the result, record: “A+ format—all three quality criteria accepted and none borderline; all mandatory conditions YES; discretionary invalidation 6c: NO.” This demonstrates the format, not authorization to trade this synthetic card: its evidence field does not contain a tested edge.

12. A similar but rejected case—synthetic circumstances, a different decision. The same pre-London window produces a nine-pip range. Price extends 11 pips beyond it and an M5 candle closes outside. The visible trigger can look much like the valid-format example, but the mandatory 12–35-pip range condition fails. Class C: no entry. The arithmetic also shows the cost effect of a narrower range. The stop distance is measured from the fill, not from the range boundary: 11 pips of extension + nine pips of range + five pips of buffer = 25 pips. The same 2.5-pip friction is now 0.10R rather than 0.07R, and each pip of slippage weighs 1.4 times as much in R terms: 35/25. Record “SKIP—range nine pips, below 12.” What price does afterward does not retrospectively change the quality of that rule-compliant rejection.

13. Evidence status. This card's status is an illustration using synthetic data—not a backtest, demo-forward record, or live history. A real evidence entry might read: “Backtest [dates, n instances, cost model]; demo forward observation [n and dates]; live record [n, start date, card version].” A candidate without its own evidence remains a hypothesis, not a setup approved for trading at a smaller size. Evidence status and opportunity class answer different questions and belong in different fields. The A+/B/C worked labels here teach classification mechanics; they do not upgrade synthetic examples into supported strategies.

Setup figures: what they do and do not establish

Synthetic reporting figures, not historical results. The following values show a checkable reporting format. The v0.1 card above is frozen before testing and has no empirical record. This separate table illustrates a hypothetical v0.2 record after a quarter. Its exit rules still impose the 1.75R planned gross ceiling, so an average realized winner cannot be assumed above that ceiling within the simplified model.

At n = 73 and a rounded 47% win rate, the approximate 95% Wilson interval is 36–58%, assuming independent observations. Event clusters can make that assumption optimistic. With a payoff ratio of 1.3, the simplified break-even win rate is 1/(1 + 1.3), approximately 43.5%. The interval includes that threshold. It therefore does not establish either a positive edge or its absence. The original example defers promotion and keeps an already authorized setup on a reduced B budget pending review; that is a governance example, not permission to start trading an unvalidated hypothesis at half size. Also separate setup approval from the A+/B/C classification of an individual opportunity. Finally, the Wilson interval addresses only win rate. Payoff has its own uncertainty, which that interval ignores. Article 15.9 covers uncertainty around expectancy itself. The number 73 is not a pass mark; the relationship between uncertainty and the decision threshold matters.

Check the drawdown arithmetic. Five consecutive 1% losses compound to roughly 4.9%, not 12%. A drawdown of approximately 12% from a losing run alone would require about thirteen consecutive 1% losses—1 − 0.99¹³ ≈ 12.2%—or five losses at approximately 2.5% risk—1 − 0.975⁵ ≈ 11.9%. Those comparisons use the same compounded convention. Actual maximum drawdown still comes from the entire equity path; interspersed winners do not guarantee that a longer drawdown has ended. If a card's number cannot be reproduced from the sizing convention and path, it is not an auditable metric. Retain older version results when classification changes so the direction of change remains visible.

One explicitly identified place for judgment

The example card is mechanical except for invalidation 6c: assessing whether an unanticipated event is driving the entry-window move. The possible event list is open-ended, so judgment cannot be eliminated simply by writing YES/NO. If a discretionary element is allowed, make it singular, named, and recorded before the decision: “discretionary assessment: YES/NO, reason.” After enough appropriately recorded observations, compare outcomes and rejected-candidate records with and without that assessment. If the discretionary component appears worse, investigate it rather than immediately discarding the whole setup. Two subgroups have their own often-wide uncertainty, and an invalidated candidate must not be turned into a live trade just to collect data. Any rule change belongs to the review process in article 15.10.

Every copy of this card needs the same qualification: it is not a demonstrated profitable system. It is a format for distinguishing a permitted setup from a similar-looking chart and later identifying what drove the decision. An edge, if present, is supported by evidence about the process—not by the template.

5. Market conditions on the card: context, calendar, and correlation

A setup does not work “always.” Any evidence for it belongs to the environment in which it was investigated, and that environment needs to be a mandatory condition on the card, not a passing comment. Traditional regime descriptions—trending, ranging, or experiencing a volatility shock—often use direction and volatility measures such as ADX and ATR^[4]. Article 15.6 covers the indicator catalog, thresholds, and validation procedure. The setup card records the particular tested definition, and the journal uses the same one.

The plan and playbook consequence is straightforward: if a regime condition is mandatory, leaving that regime disables new entries under the setup. There is no vote and no “one more try.” A setup that once worked and now does not may have left its tested environment rather than permanently lost its edge; alternatively, the evidence may never have been robust. Article 15.10 addresses how to distinguish those possibilities. The card should make the boundary explicit before that diagnosis becomes necessary.

The calendar clause

Every card defines an event window in which it does not permit entry. This article's example uses ±2 hours around relevant high-impact announcements, with the stricter flat-before-event requirement where specified. A separately approved event strategy needs its own mandate, card, and sizing; it is not an improvised exception. Review the coming week and check the calendar service's time-zone setting. “Thirty minutes before” applied in the wrong zone is not the intended rule. Recheck changes and unexpected announcements as required by the plan.

Identify excluded days by event name, not a memorized position in the month, because schedules can change. The US Employment Situation report, commonly called the NFP release, is normally scheduled for 08:30 America/New_York; verify the date in the Bureau of Labor Statistics' current calendar rather than assuming every release is on the first Friday. A scheduled FOMC decision normally follows the second meeting day at 14:00 America/New_York; consult the Federal Reserve's calendar and announcement. Those correspond to 14:30 and 20:00 in Poland during much of the year, with an hour's difference during mismatched daylight-saving weeks. Record the decision day, not the following day. The card can say: “Excluded: Employment Situation release day and FOMC decision day; dates checked against the publishers' current-year calendars.”

The correlation clause

Two positions each carrying 1% planned stop risk total 2% planned loss if both stops execute at the assumed prices, before unmodeled costs or slippage. Correlation does not reduce that sum. Similar exposure can make losses more likely to occur in the same move, which is why the filter exists. A practical rule is: “If another position already carries the same prohibited currency exposure, do not open this one; where the plan explicitly permits it, allocate one shared risk budget.” The London example's card chooses no additional same-exposure position, so a generic shared-budget option does not override it. Article 15.6 covers measurement windows and thresholds; the card records the applicable definition rather than inventing a new one at entry.

When the playbook must fit your life calendar

A playbook that cannot be executed alongside your job is a wish list. If a setup requires live observation during hours when you cannot provide it, either a tested, explicitly specified pending-order or alert workflow makes it feasible, or it does not belong in your approved playbook. Replacing a confirmation entry with a pending order changes the strategy and requires testing; the substitution is not automatic. Article 15.5 covers daily and weekly schedules, including employed traders. In the plan, the decision belongs in two places: section 4, available time, and section 2, the mandate.

An accepted server-side stop is part of making this card operationally feasible. A mental stop cannot act while you are disconnected or absent. It still does not guarantee a fill price, and local trailing logic may need a running platform. Article 15.1 covers workstation dependencies and redundancy; article 15.8 covers an interruption while exposure is open.

6. Entry, exit, and the last risk check before submission

A rule is unambiguous when someone else can apply it

The test is simple: give the card to someone unfamiliar with your personal chart-reading habits and ask whether they reach the same decision from the same information. “Enter when the market shows strength” fails. “Enter after an M15 candle closes above the specified previous high” is much more testable, provided “previous high” is defined. Article 15.3 audits each strategy rule into its component, definition, ambiguity, and correction. For the playbook, the consequence is that every card statement is either an observable condition or an explicitly identified judgment. An unnamed interpretation gap is not a third legitimate category.

Stop loss: the price condition that invalidates the trade premise

The stop loss belongs at the level defined to invalidate the particular trade premise, not at whatever distance lets you buy the preferred size. For the hypothesis that a pre-London breakout will continue, returning beyond the opposite side of the range is the example's price invalidation. It is a rule for this trade, not proof that the whole strategy is false after one stop. Field 12 shows the cost implication: a narrow range can mean a shorter entry-to-stop distance, making the same transaction costs larger in R terms. That is why a range below the threshold is rejected here rather than celebrated as a chance for a tighter stop. Size follows the specified stop and budget; the stop is not moved to justify size.

Take profit and partial exits

A take profit can be a fixed R target—1R, 2R, or 2.5R—or a defined structural price level. Fixed targets are easier to execute consistently and compare in the journal. Structural targets may perform differently in particular conditions, but can add decisions and ambiguity. Which works better for your setup is an evidence question. Compare specified versions, not a mixture in which the exit was chosen after seeing each trade develop.

Partial closure, such as half at 1R and the remainder farther away, is a common template rather than a universal standard. It realizes some profit earlier and may make it easier for a particular trader to follow the remaining rule. It also reduces the size participating in a larger move and complicates comparison with a single full-position target. With only a dozen or so trades in a quarter, uncertainty can be wide, and changing the rule creates a new version-specific sample. Do not move the stop to entry after +1R or adopt partial TP merely because others do. If many successful trades naturally retest the entry after +1R, that stop move may remove them before the larger move. Investigate your own data and record an accepted change as a new card version.

Maximum excursions help investigate stop and target placement: Maximum Adverse Excursion is the worst marked outcome during the trade, and Maximum Favorable Excursion the best^[3]. They can reveal questions about stops that are too close or targets that are poorly matched, but they do not by themselves validate an optimized exit. Article 15.9 gives the definitions, calculations, scatter plot, and diagnostic table. Here the output of the analysis is a specific exit-rule proposal, tested and, if accepted, recorded with a new version and date.

The risk filter: the final gate before clicking

The example allocates about 30 seconds to the final risk check. It revisits the mandatory gate in a fixed order; the target time does not justify skipping a check:

1. Has the monthly realized result reached −5% of month-opening capital? Apply the plan's half-size regime. At −8%, stop for the month. Confirm that daily and weekly stops have not already prohibited trading.
2. Would this entry take aggregate planned open risk above 3% of the defined capital base? If yes, no entry.
3. Is there a relevant high-impact release within the applicable ±2-hour window? If yes, no entry under this card.
4. Does the new position duplicate prohibited currency exposure? Decline it, or use a shared budget only where the governing card and plan explicitly allow that alternative.
5. Does the approved card have an ID and version, and do all mandatory conditions pass? Any NO means rejection and a recorded reason, not a smaller C-class trade.

Without this gate, someone who says “I risk only 1% per trade” can accumulate 4–5% of planned risk across related positions and discover the concentration only when multiple stops are triggered by the same move. The individual percentage never replaced the aggregate exposure check.

7. Hard rules, interpretive guidance, hypotheses—and what to leave out

Not everything written about a setup has the same status. Mixing rules, guidance, and hypotheses makes it difficult months later to tell what was breached and what was merely considered. Give each statement an explicit status:

The operational playbook contains hard rules and explicitly named guidance. Hypotheses have a separate record and research rhythm. Article 15.3 explains how a research question becomes a specification with rejection criteria.

Leave out motivational slogans

“I trade with discipline” and “Patience is the key to success” are poster statements, not executable rules. Consistency is supported by the procedures in plan sections 3–8, not by repeating an aspiration. A slogan occupies space that could hold a condition you can actually check.

Do not present untested ideas as approved setups

A card without its own evidence honestly says “hypothesis” in field 13 and belongs among research candidates. No fixed trade count automatically promotes it. State n, uncertainty, and independence assumptions; relate the evidence to the setup's decision threshold. If uncertainty spans the threshold, the review decision is unresolved. For a previously approved setup, the governance process may maintain or reduce an existing allocation, or pause it, depending on the plan and evidence. That is different from authorizing a new hypothesis as B merely because its uncertainty is wide. An operational problem—a breached limit, execution fault, or outage—can justify an immediate pause without waiting for a statistical sample.

Leave out internet setups you do not understand

A chart with a dozen indicators shown by an influencer does not belong in your approved playbook until the relevant rules are understood and tested. “It worked for someone” is selected evidence, not proof of a transferable edge^[6]. The same nominal rules can produce different net outcomes with another feed, provider, cost structure, or operator. Field 9 is where your cost model makes those differences explicit.

Do not create a second set of plan rules

Sizing limits, daily stops, allowed hours, and session-stop conditions belong to the plan. A card can quote or reference them—for example in field 8—but cannot establish conflicting numbers. Independent copies create two sources of authority; after the first change, a contradiction may remain hidden until an incident.

Leave out unsupported macro forecasts

“The Fed will probably cut rates” is not a setup. It is a hypothesis or contextual view. The playbook needs a testable entry and exit process for the relevant conditions, plus an invalidation rule. Even if the macro view proves right, the specified trade is not guaranteed to profit; price, timing, execution, and prior expectations still matter.

8. Versioning and review: when a card may change

A playbook is not written once and forgotten, but it is not rewritten after every loss either. The difference is explicit: a change happens at an authorized review or predefined trigger and receives a version number. An undocumented response to the latest result is not calibration. It can destroy the comparability of the sample you have spent months collecting.

Step 1: collect the figures

From the journal in article 15.9, extract each card version's closed-instance count, win rate, realized payoff, net expectancy and uncertainty, longest losing run, setup drawdown, and share of trades with rule breaches. Article 15.9 owns the formulas, denominators, and interval methods; here you collect the measures needed for a decision about the card.

The prerequisite is that records are reconciled to the provider's statement and identified by version. A pooled result from three changed specifications cannot be treated as the result of any one of them. Version mixing is one possible explanation for a setup that appears suddenly to have stopped working.

Step 2: decide the card's status and permitted allocation

Article 15.10 owns the evidence criteria for retention, observation, downgrade, upgrade, and suspension. Three principles concern the card itself:

• An interval, not just a point. If expectancy uncertainty spans the decision threshold, the statistical conclusion is deferred. The existing approved setup may remain under observation, have its allocation reduced, or be paused according to the governing plan; it should not be declared dead merely from one weak quarter. This does not authorize an untested candidate.
• Check execution before blaming the strategy. Reconcile the data, apply the cost convention from field 9 consistently, and identify trades that breached mandatory conditions. A series of noncompliant trades is not a clean test of the specified setup, even though those losses still belong in the account record.
• A changed allocation rule needs a new version. If the review changes permitted sizing—for example from the full to the reduced budget—update field 8 and the version in field 1. Keep this card-level governance decision distinct from classifying the next individual opportunity. Without versioning, later records cannot show which sizing rule applied.

Step 3: implement and archive

Save the new card, archive the previous version, and add a change-log entry with the version, date, changed rule, and one-sentence reason. After a year, that becomes the history of the setup's development. Without it, even two quarters later you may be unable to explain why a threshold has its current value and may end up repeating old research from the beginning.

Step 4: maintain the graveyard

Keep suspended or removed setups in a separate “graveyard,” with the last version, reason for removal, and conditions under which renewed research would be worthwhile. Otherwise, months later you may rediscover an idea you already tested unsuccessfully. For example: “PINBAR-D1 v1.3—removed after two consecutive reviews with negative net expectancy. Reason: no regime condition on the card; the historical behavior was confined to elevated volatility. Revisit only after defining a regime gate and conducting a fresh test as v2.0.” This is an illustrative archive entry, not evidence that the proposed regime repair will work.

9. Common mistakes when writing the plan and playbook

Mistake 1: a twelve-page plan no one reads

The one- to three-page target concerns the working plan, not the length of this explanatory lesson or its supporting research. Eight or twelve pages of unstructured instructions can disappear from the daily workflow because no one consults them before the session. Brevity is an operational property here: remove repetition, not required conditions. If your plan is six pages, check whether the same rule appears in three different places and whether detail belongs on an individual setup card or in supporting documentation.

Mistake 2: a playbook without figures or versions

“D1 pin-bar reversal” without an instance count, expectancy definition, or the version those figures describe is a note, not an auditable operating record. Field 13 must be honest, even if it says “hypothesis—no evidence yet.” Writing a candidate card is allowed. Presenting it as a validated A+ setup, or trading it merely because the label is attractive, is the problem.

Mistake 3: failing to review changed circumstances

A plan written for a PLN 10,000 account needs review if the account becomes PLN 50,000. The same 1% is PLN 100 in the first case and PLN 500 in the second; a larger monetary loss may affect behavior differently even though the percentage has not changed. Those Polish amounts are examples of scale, not deposit recommendations. The same reasoning applies in another currency. Review the whole plan annually and the playbook quarterly under this example's schedule, with an earlier review for a defined material change in capital, available time, or other governing conditions.

Mistake 4: no STOP conditions

Without an explicit stopping rule, one loss can start a sequence: an urge to recover, more tension after the next loss, and another trade intended to fix both. The damage may come from several entries after the trader's own limit should already have ended the session. STOP conditions must be hard, triggered on reaching the defined limit; implementable, through an understood provider control where available or a documented procedure for positions, pending orders, logout, and recording; and stable, not renegotiated while the session is in progress. A rule held only as an intention is easier to override than an observable trigger with a prepared action. Closing the terminal alone is not a position-closing procedure.

Mistake 5: two names for the same setup

“London-open breakout” and “London-open range break” may be the same process under different names, splitting the sample without adding a genuine distinction. Compare the mandatory conditions side by side. If only one threshold differs, document whether you are studying two versions of one setup rather than two unrelated strategies. Do not pool the versions as though they were identical, but do not give duplicate labels the appearance of diversification either.

Mistake 6: no record of what was removed

Keep the graveyard described in section 8. Consider a major-pair D1 pin-bar reversal removed after two adverse reviews, with the note “no regime gate; favorable historical results confined to elevated volatility.” A year later, the trader encounters a newly promoted pin-bar setup online. The archive reveals the previous work. Instead of restarting the same incomplete specification, the trader can investigate a clearly versioned regime hypothesis with a fresh test. A short record of what was removed and why can prevent months of repeating the same mistakes. It does not make the revived candidate automatically valid.

10. Copyable templates: the plan and the setup card

Open your preferred document tool or spreadsheet and create two records: “Trading Plan v1.0” and “Playbook v1.0.” Use the structures below to produce a first working draft. One afternoon may be enough to expose the missing decisions, although completing the supporting research takes longer. The first version does not need to look elegant. It needs a structure that makes the next period of observation and any independently authorized trading interpretable. Do not trade an unvalidated hypothesis simply to fill the journal. At the next quarterly review, you may discover that the proposed hours were unrealistic because you worked when you were free rather than when the setup was defined to operate.

Trading-plan template: nine sections

TRADING PLAN — version [X.Y], effective from [YYYY-MM-DD]

1. PROCESS OBJECTIVE (first 6–12 months)
   No daily or weekly limit breaches
   Every trade and rejection recorded before its outcome is known
   No entries outside the approved playbook
   Outcome scenario: [add only after the process objective is established;
   state assumptions and arithmetic, not a forecast or return quota]

2. MANDATE AND SCOPE
   Market, account product, and instruments: [...]
   Horizon and maximum holding time: [...]
   Working method: [manual / permitted pending orders / automation]
   Approved strategies: playbook v[X.Y], cards [ID v..., ID v...]
   Anything outside this list: no trade

3. RISK LIMITS
   Per A+ trade: [%] | Per B trade: [%]
   Aggregate planned open risk: maximum [%]
   Daily realized-loss limit: [%] (= [N] full A+ losses on the stated base)
   Weekly limit: [%]
   Monthly: [%] → halve sizing; [%] → STOP and review
   Period-limit denominator: capital at the start of that period
   Measure: realized result; open exposure tracked separately
   Individual trade-sizing denominator: [define explicitly]
   REACHING the threshold triggers the rule; no overshoot is required
   Planned stop risk is not a guaranteed maximum loss

4. AVAILABLE TIME AND TRADING HOURS
   Available time: [N] minutes on working days, [N] at the weekend
   Active entries: [hours and named time zone]
   Flat: [hours and named time zone]
   Mandatory event exclusion: [card's window; this example: ±2 hours]
   Pending orders: [activation window and cancellation/expiry rule]
   Rollover: [provider's convention and event time zone]
   Local conversion: [check daylight-saving differences for the date]

5. ENTRY PROCEDURE: FIVE CHECKS BEFORE SUBMISSION
   1) Identify the approved card and version
   2) Check mandatory regime, calendar, correlation, and limit conditions
   3) Calculate size using the stated model
   4) Prepare the order ticket, SL, and TP before confirmation
   5) Prepare the decision record before the outcome
   After submission: add acknowledgment, actual fill, and order IDs

6. OPEN-POSITION MANAGEMENT
   After +1R: [card rule, supported by the relevant analysis]
   After +2R: [card rule, if applicable]
   SL only moves in the risk-reducing direction; never widened
   TP changes only under a condition specified before entry

7. STOP CONDITIONS
   Two consecutive losses: 60-minute pause in this example
   Daily result at [-%]: end the session
   Three consecutive B losses: end the session in this example
   Weekly limit: pause until the next permitted Monday
   Provider-side control: [available / unavailable; exact behavior]
   Procedural alternative: [handle positions, cancel unwanted pending orders,
   then log out and record; closing the terminal closes no server-held order]
   Resume: after the required pause and documented process review,
   NOT after recovering the loss
   Stricter stop conditions override shorter pauses

8. BEHAVIORAL RESTRICTIONS AND NO-TRADE RULES
   [Rule] | [observable trigger] | [action] | [permitted exception]
   | [record location] | [authorized change owner]
   An exception outside the approved list is not permitted

9. RECORDS, REVIEW, AND CHANGE CONTROL
   Every trade and rejected candidate: decision recorded before outcome
   Daily: journal | Weekly: rule-compliance review
   Monthly: metrics | Quarterly: playbook review
   Annually: whole-plan review
   Changes only at review or after a predefined material trigger
   Each accepted change: new version and one-sentence rationale
   Previous version: archive, do not overwrite
   Immediate pausing for an operational or safety problem remains allowed

Setup-card template: thirteen fields

1. ID AND VERSION: [SETUP-CODE vX.Y], effective from [YYYY-MM-DD]
   Previous version: [vX.Y] archived | Change: [one sentence]

2. HYPOTHESIS: [proposed market behavior and mechanism]
   Rejection criteria: [evidence that would contradict the hypothesis]

3. CONTEXT: regime [...], session [hours and time zone], instrument [...]

4. MANDATORY CONDITIONS: all YES; missing data means NO
   - [...]
   - [...]
   - plan limits permit entry; order ticket complete

5. TRIGGER: [specific observable condition and timeframe]

6. INVALIDATION: [what disqualifies the candidate before entry]

7. ENTRY AND EXIT
   Entry: [one specified order method; alternatives need separate versions]
   SL: [defined price condition, not a preferred monetary loss]
   Stop after TP1: [...]
   TP1: [...] | TP2: [...]
   Trailing: [activation condition, update timing, and exit event]
   Planned gross ceiling implied by exits: [x]R, if a ceiling applies
   Same-bar ambiguity in tests: [explicit convention]
   Time exit: [deadline and named time zone]

8. SIZING MODEL AND QUALITY CRITERIA
   A+ [%] | B [%] | C: no entry
   Sizing quoted from plan section 3, not independently invented here
   Quality criteria, each with its threshold and permitted-side border band:
   (1) [name]: accepted threshold [...] | borderline [...]
   (2) [name]: accepted threshold [...] | borderline [...]
   (3) [name]: accepted threshold [...] | borderline [...]
   This example's convention: within 5% on the accepted side only
   Specify strict versus inclusive inequalities and any rounding
   Mandatory failure or invalidation overrides every quality class

9. COSTS
   Friction reflected in fills: spread [x] + slippage [x] = [x] pips
   Equivalent [x]R at an initial stop distance of [x] pips
   If already included in fills, do not deduct it twice
   Additional net deductions outside R0: commission [x] + financing [x]
   R0 convention: [exactly what is included and excluded]

10. PROHIBITIONS: [actions not allowed under this setup]
    Excluded days: [event names, not assumed positions in the month]

11. VALID-FORMAT EXAMPLE
    [Conditions, trigger, entry, SL, TP, and class recorded before outcome]

12. SIMILAR BUT REJECTED
    [Similar chart, failed mandatory condition, and journal rejection entry]

13. EVIDENCE STATUS
    [Hypothesis / backtest: dates, n, costs / demo forward: n and dates /
    live record: n, dates, provider, and card version]
    A hypothetical illustration is not trading authorization

STATISTICS: version [X.Y], n = [N]
   WR: [%] | Realized payoff: [x] | Net expectancy: [±x.xxR]
   95% WR interval: [x–y%], method [...], independence assumptions [...]
   WR uncertainty does not include uncertainty in payoff
   Expectancy uncertainty: [method, interval, and assumptions]
   Longest losing run: [N]
   Drawdown: [%] from the full defined equity path, not only the longest run

JUDGMENT ELEMENT: [one named assessment, recorded before the decision]
GOVERNANCE STATUS: [approved / observation / suspended / research only]
Keep governance status distinct from the class of an individual opportunity