How Often a Plan Should Be Allowed to Change
Two failure modes sit at opposite ends of the same question. One trader has carried the same document for years and defends it against every piece of contrary evidence. Another rewrites theirs after any session that disappoints. Both are avoiding the same difficulty, which is that a written plan is a claim about how a strategy should be run, and claims have to be revisable without being disposable.
The Timing of the Edit Decides Whether It Is Legitimate
The single most useful discipline is separating the moment of revision from the moment of trading. A change made on a weekend, with the week's records in front of you and no position open, is a revision. The identical change made at ten in the morning with a trade going against you is an improvisation, and the fact that it was typed into the document does not launder it.
This is worth stating in the plan itself. A line specifying that the document may not be edited during market hours costs nothing to include and removes the most damaging category of change entirely. It also makes the rule visible, so that when you find yourself wanting to edit mid session you at least know what you are doing.
Reviews on a Schedule, Not on a Feeling
Fixing a review interval solves most of the frequency problem by itself. A short weekly pass that only asks whether the rules were followed, and a longer periodic pass that asks whether the rules are any good, keeps the two questions apart. Conflating them is how a trader ends up changing a sound rule because they broke it, which is the wrong repair for that particular fault.
Between reviews, observations go into a list rather than into the document. Most of them will look less compelling a week later, which is exactly the filter you want. The ones that still look compelling have survived the emotional context that generated them, and those are the candidates worth considering.
How Much Evidence a Change Should Require
The honest answer is more than feels necessary. A single session tells you almost nothing about a rule, because the range of normal outcomes for any strategy comfortably contains sessions that look like the rule failed. Changing after one is fitting the document to noise, and doing that repeatedly produces a plan that describes recent history rather than a method.
A reasonable bar is that the same problem has appeared repeatedly, in circumstances you can describe, and that you can state what the change is supposed to fix before you make it. If you cannot articulate the failure the edit addresses, you are not improving the plan. You are relieving discomfort.
Some Changes Are Not Really Changes
It helps to distinguish between amending a rule and clarifying one. Discovering that your plan never specified whether the range is measured on bodies or wicks, and writing down the answer you have been using all along, is not a revision. It is closing a gap, and it should be done immediately rather than waiting for a review, because the gap is where discretion leaks in.
Genuine amendments change what you would do in a situation you have already faced. Those deserve the full process. Clarifications change nothing about your behaviour and only make the document match it, and treating the two the same way makes people reluctant to fix obvious holes.
Keep the Old Versions
A plan with no history is difficult to learn from, because you lose the ability to ask when a rule arrived and what it replaced. Dating each version and keeping the previous ones, with a line explaining why the change was made, turns the document into a record of your own reasoning rather than a snapshot of current opinion.
The value shows up later, usually when a rule starts to feel arbitrary and the temptation is to drop it. Being able to read why it was added, and what went wrong before it existed, is often enough to answer the question. It also exposes the pattern where a rule is removed, reinstated after a bad stretch, and removed again, which is a cycle that is nearly invisible without a version history and obvious with one.