Write It Before You Trade It

Five fields, one of them arithmetic

A plan is a short written statement that answers five questions. What is the setup — the repeatable condition that makes this interesting, described so a stranger could check it? What is the entry trigger — the specific event that converts interest into an order, usually a close or a reclaim rather than a price touch? Where is the stop, stated as the price at which the idea is disproved rather than the price at which the loss annoys you? What is the target, and is it a level with a reason or a trailing rule? And how large is the position, given that the risk budget is a fixed fraction of capital and the distance from entry to stop is the divisor? The first four are judgement about the market; the fifth is arithmetic, and it is the field beginners leave out, which is why a plan can be right about direction and still lose money. Once those five exist, the trade can be expressed in R — units of initial risk — and that is what makes a sample judgeable. A 2.5R winner and a 1R loss are the same units regardless of account size, so ten trades produce an expectancy figure that does not depend on how much capital was in play, and a journal can compare setups across months without inflation or account growth distorting the record. The arithmetic also sets the bar for a win rate: at a 2:1 payoff the break-even win rate is 33%, at 3:1 it is 25%, and those numbers are checkable before the first trade rather than after the tenth. The plan’s other job is to move decisions out of the moment. Every field that is written in advance is a decision that does not have to be made while the position is live, and the fields that matter most are the stop and the size — the two a learner is most tempted to adjust once the price moves. That is also why the plan is short: a page of context is a document nobody reads in the ten seconds before an order, while five lines can be checked. Pre-commitment works because the moment of action is the worst time to be deciding the terms of the action, and the written form is what makes the pre-commitment real rather than intended. One plan, five fields, one arithmetic line — Setup: daily close above the 50-day after a three-week base, volume ≥1.5× average: The condition, checkable by a stranger · Entry $50.00, stop $48.00 at the base low, target $56.00 at the prior high: Risk $2.00, reward $6.00 ← · Risk budget: 1% of $80,000 = $800: Size = 800 ÷ 2.00 = 400 shares · Expectancy at 45% wins: 0.45 × 3R − 0.55 × 1R = +0.80R a trade ← The plan is not a forecast. It says what you will do under each outcome; the market decides which outcome arrives. A plan that needs the market to cooperate to be a plan is a wish with a price target.

How a plan fails, and how to tell which way

Three failures look identical from the outside and need completely different responses. A plan can be executed badly: the fields existed and were not followed, which is the only failure that is immediately yours to fix. A plan can be designed badly: the stop sits inside the noise from TR3, the target is inside the spread, the setup fires in the range regime from TR14 — the arithmetic was applied to an idea the market does not pay for. And a plan can be unlucky: the edge is real, the sample is small, and a run of losses arrived. The first is a discipline question, the second is a design question, the third is only answerable in retrospect, and the tool that separates them is the journal with the setup recorded. There is a fourth failure with no cure but worth naming because it dominates the others in practice: a plan that is technically complete and emotionally unusable. A stop that requires accepting a loss the trader cannot take is a stop that will be moved, and a size that produces a sleepless week is a size that will be cut at the worst moment. The fix is not more resolve; it is making the parameters survivable — a smaller risk fraction, a smaller position, or simply fewer trades — because the plan has to be one that can be followed on the worst day rather than the best one. So the mastery test for this lesson is not whether a plan can be written. It is whether the plan survives contact: the stop was where it was written, the size was what the arithmetic said, and the next entry records the same five fields. The last one is the loop — the journal in TR21 — because a plan that is never audited is a ritual rather than a process, and the audit is what turns four weeks of trades into evidence about whether the setup is worth trading at all. • Bad execution: the fields existed and were not followed — fix immediately. • Bad design: the stop sat inside the noise, or the setup fired in the wrong regime. • Bad luck: a real edge with a small sample, only answerable in retrospect. • Unusable plan: parameters the trader cannot live with, which guarantees the stop gets moved. • The journal is the audit that separates the three curable failures. Do not confuse a plan with a prediction. A plan states what you do under each outcome and how much it costs if the outcome is bad; it does not need to be right. Reading a written plan as a forecast is how a losing trade becomes an argument with the market.

The plan is the hypothesis

The five fields describe a position, but they also describe an experiment. Writing entry, stop, target, size and reason before the trade is a form of pre-registration: it fixes the conditions under which the idea is right or wrong *before* the outcome is known, which is the only time the judgment is honest. Psychological research on hypothesis testing is unambiguous that people reframe a failed prediction after the fact — the plan, written first, removes the room to do it. What that buys you is a dataset. If every trade carries a written reason, a written invalidation and a recorded outcome, then after fifty trades you can ask questions a P&L cannot answer: which *reasons* had an edge, which exits gave back the most, and where the size was least justified. None of those is answerable from a list of wins and losses, because profit and loss is one number and the plan is five. The last field is the one most people skip, and it is the load-bearing one. A target says what you hope will happen; an invalidation says what would prove you wrong. If the plan contains only a target, every adverse move becomes a reason to wait, and the stop becomes a number you negotiate with. If it contains an invalidation, the exit is decided in advance by the same person who entered, in a calmer state. A plan with no invalidation is not a plan — it is a preference. The test is whether you can write the sentence “this trade is wrong if…” without reference to how much money you are down.

A plan you do not read is a wish

The five fields and the arithmetic line are the content of a plan, and the content is not where plans fail. They fail operationally: they are written once, filed, and then reconstructed from memory during the trade, at which point the reconstruction is shaped by the position’s result so far. What turns a written plan into an operating one is placement and routine, and both are smaller than they sound. Placement first. A plan has to be visible at the moment the decision is made — on the screen beside the order ticket, not in a document. The practical reason is that the plan’s whole function is to be consulted before the click, and anything that requires navigating away from the order is not consulted under pressure. The same applies to the exit: the stop and the target belong in the broker’s system as orders, because an exit held only in the plan is an exit that has to be remembered. Then the routine. A pre-trade checklist run out loud — setup, trigger, stop, size, target, and what would disprove the idea — takes under a minute, and its value is that a blank field stops the trade. That is the specific mechanic worth borrowing from aviation: the checklist is not there to improve the good decisions, it is there to catch the decision where one input is missing and the omission is invisible to the person making it. The failure modes to name are the ones that arrive disguised as judgement. **Plan drift**: the stop moves because the chart now looks fine, which is a decision made with the outcome partly known. **The retro-fitted plan**: a rationale written after the entry and given the language of a plan, which is why a plan is worth much less if it was not timestamped before the order. And the **outcome rather than condition** error: “I will exit when it starts to look weak” specifies a feeling, not an observation, so it can never be checked and can always be satisfied either way. The audit is the same one this curriculum uses elsewhere and it takes ten minutes: take the last twenty entries and check each against its plan. Trades that fall outside the specification are either evidence against the rule or evidence that the rule was not followed, and the two findings need different responses. There is also a category people rarely log — the plan that was abandoned and the trade that still made money — and it is the most instructive of all, because it is the case where the process quietly lost to luck and the lesson extracted will be the wrong one unless it is written down. • The plan has to be visible where the order is placed, or it will not be consulted. • Working exits belong in the broker’s system, not in a memory of the plan. • A checklist that stops a trade on a blank field is the point of the checklist. • Log the abandoned plan that still won; it is the case that teaches the wrong lesson otherwise. The smallest version that works: before every order, one line — setup, stop, size, and the thing that would prove the idea wrong. If a trade cannot be described in that line, it has not been planned, regardless of how much has been thought about it.

Sources

• Trading plans and process disciplineElder, “Trading for a Living”
• R multiples, expectancy and position sizingTharp, “Trade Your Way to Financial Freedom”
• Pre-commitment and implementation intentionsGollwitzer (1999) on implementation intentions; behavioural finance literature