Trading Plan: Written Down or It Does Not Exist
A trading plan is a written specification of what you will trade, under what conditions, at what size, and when you will stop. Writing it down is what makes it a plan rather than a preference, because an unwritten rule adjusts itself to whatever just happened.
How it works
A plan is a document. Not an intention, not a general approach, and not something you could describe if asked. Something written, dated, and specific enough that a stranger reading it would take the same trades you would.
Four questions cover most of it. What instruments, in what conditions, at what size, and with what exit. Everything else is elaboration.
The writing is not administrative. An unwritten rule adjusts silently to the last outcome — after two losses it becomes more cautious, after two wins more generous — and the adjustment is invisible because there is nothing to compare against.
The part that actually decides the result
Position size moves results by more than every other decision combined. The same setups at half the size produce half the swings; at twice the size they produce an account that can be ended by an ordinary losing run.
Entry technique is the part everybody works on and the part that matters least. That is not a popular claim and it follows directly from the arithmetic: a small change in edge moves the outcome slightly, and a doubling of size doubles every deviation.
A plan that does not permit a blank day produces trades on days that had none. Writing “no trade is a valid outcome” sounds trivial and removes a specific pressure that otherwise operates every session.
The hours available to you are a constraint on the plan, not a detail. A method requiring attention during hours you are at work is not a method you have; the honest version is to choose an approach that fits the time you actually have.
In practice
Set a review date and change nothing before it. Without one, a plan is revised after every uncomfortable outcome, which means it is never tested — and a plan that is never tested cannot be improved, only replaced.
Monthly or quarterly are both defensible. What matters is that the date is fixed in advance and that changes happen on it rather than in response to the most recent loss.
Every trade costs 2% of a median bar’s range in round-trip costs on this site’s shared history — and 45% of the smallest bar in the series. Which makes intended frequency a part of the plan rather than an outcome of it: a method taking four trades a day is paying that cost a thousand times a year, and the plan should state the number rather than discover it.
Volume decides which hours are worth trading, because a fixed cost against a small bar consumes most of the available move.
A gap overrides the stop rule entirely, which is why overnight position size has to be a separate decision from intraday position size.
A useful test for whether a plan is specific enough: hand it to somebody who does not trade and ask them to identify today’s trades. If they cannot, the plan contains judgement it has not written down, and that judgement is where the outcome is actually being decided.
The same test applies to the exit. “Take profit at a sensible level” fails it; “exit at two times the initial stop distance, or on a close below the 20-period average, whichever comes first” passes. The second is not obviously better as a method — it is better as a plan, because it can be followed and it can be shown to be wrong.
And a plan should say what would make you stop trading it entirely. A maximum drawdown, a number of consecutive losing months, a change in the market it was designed for. Without that line, abandoning a method is always an emotional decision made at the worst moment, and with it the decision was made in advance by somebody calmer.
What a trading plan is not
It is not a strategy. A strategy is one component of it.
It is not fixed forever. It is fixed until the review date.
It is not a prediction. It specifies behaviour, not outcomes.
And it is not a plan if it is not written, which is the only claim on this page worth arguing about.
When it fails
In a range a working plan still loses slowly. It produces the trades it was designed to produce, in conditions where those trades do not work, and every one of them pays the round trip. That is a regime problem rather than a plan problem, and telling them apart is what the review date is for.
The second failure is revision after loss. The plan changes, the sample restarts, and nothing is ever measured across enough trades to say anything.
A third is vagueness. “Trade with the trend” is not executable; “long only when the 50-period average is rising on the daily chart” is.
A fourth is omitting the frequency. A plan without an expected number of trades cannot be costed.
And a fifth is a plan you cannot actually follow given your hours, your capital and your temperament — which is the most common failure of all, and the only one that is knowable before starting.
The original data
On this site’s shared 576-bar history the round-trip cost is 0.0098 price units — 2% of the median bar
range of 0.493 and 45% of the smallest bar of 0.022 — and bar ranges span 0.17 to 1.10 between the tenth
and ninetieth percentiles. The figures are in research/series-measurements.json, produced by
site/measure_series.py.
Those figures turn a plan from a document into an arithmetic problem, which is an improvement. Multiply your intended trades per year by your real round-trip cost, express it as a share of the average bar you are trying to capture, and you have the return the plan must produce before it breaks even. That number is knowable on the day you write the plan, it does not depend on any view about markets, and it disqualifies a substantial fraction of the methods people start with.
Related
Trading rules covers how to write the individual rules so they are testable. Risk per trade is the sizing decision that dominates the result. And trade review is what turns the plan into something that improves.
My first plan was four pages and I could not have told you what any of it committed me to. The version that worked is one page with numbers on it - instrument, session, size, stop, and a monthly review date - and the specificity is the entire difference.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.