Documenting And Reviewing Your Trading Plan Over Time

Refinement or rewrite, and the sample size that decides

Not every change carries the same burden of proof, and pretending otherwise leads either to paralysis or to weekly rewrites.

• Under 30 trades: nothing changes except admin. Fixing an ambiguous sentence, adding a missing definition, tightening wording so a rule is checkable. Those are clarifications, not revisions.
• 30 to 100 trades: one parameter, once, with the before and after recorded. Stop multiple, one filter, one session window.
• Over 100 trades: structural changes, such as dropping a setup type or adding an instrument, and still one at a time. Two simultaneous edits produce a result neither can be credited with.

The reason for the floor is that short runs say very little. A losing week is typically four or five trades, and at a 45% win rate the chance of four or more losses in five trades is about 26%, roughly one week in four. Any process that revises a plan after a bad week revises it about thirteen times a year, on noise. Comparing strategy families is a useful exercise at the 100-trade mark and a waste of an evening at ten.

An hour a month, an afternoon a quarter

The monthly hour is mechanical and asks four questions. What was compliance? What did costs take, in spread, swap and slippage, as a share of gross? Which setup type carried the month, and which lost? And did anything happen that the plan does not cover, which is how the stand-aside list grows. Total costs are worth pulling into money rather than pips, and if you trade enough volume the rebate calculator shows what part of it is recoverable.

The quarterly afternoon asks the harder question, which is whether the strategy is still doing what it was built to do, on a sample large enough to answer. Sixty trades is a reasonable minimum. Compare expectancy per trade against the backtested figure, check whether the market conditions over the quarter were the conditions the strategy was tested for, and look at maximum drawdown against the ceiling in the plan. The drawdown calculator turns the equity curve into the number the ceiling is written in.

The four ways a review quietly stops being a review

• It happens after good months. A review scheduled by mood is a review of wins. Fix the date, not the trigger.
• It reads the profit and loss instead of the process. Profit and loss over 30 trades is mostly variance. Compliance, costs and setup mix are readable at that sample; the edge is not.
• It changes two things. Both changes then have to be reversed together, and the next quarter cannot attribute anything.
• The document is never opened. If you have not read your own plan in eight weeks you are not trading it, you are trading a memory of it, and memories drift towards whatever you did last.

End every review by writing the next review date in the document before you close it. It is the one line that makes the whole thing recur without anyone deciding to.