A trading plan is a written statement of what you will do, decided before the market gives you a reason to do something else. For a discretionary trader it covers entries and exits. For someone running an EA, the entries and exits are the EA's job, and the plan covers everything around it: which system, what risk, where it runs, how it is judged, and when it stops.
What to write down
The EA, its version, and the settings, saved as a file so they can be restored exactly. The risk setting and the reasoning for it, usually the worst drawdown on the record that you would sit through. The broker, the account type, and the minimum balance. The VPS and how it is checked. The markets the EA is allowed to trade and any it is not.
Then the review: how often you will look at the record, what you will look at, and what you will not act on. A weekly check of the Experts log for errors and a monthly review of the equity curve, drawdown and trade count is sufficient for most systems. A daily inspection of profit and loss is not a review; it is an invitation to interfere.
The rule for stopping
The most important line in the plan is the one that says when the EA will be switched off, written while you are calm. The right trigger is a drawdown beyond what the record supports, say the DD95 or the maximum on record plus a margin, or a change in behaviour: trades that do not match the description, stops missing, error messages, or a broker change. The wrong trigger is a losing month, which every strategy has, and which is the moment most owners switch off and lock in the loss.
Equally, write down what will not cause you to intervene: a losing week, a single large loss at the stop, a period of no trades, a profitable month that tempts you to raise the risk.
A trader's plan states: run the EA at the medium setting; switch off if the equity drawdown exceeds 22 per cent, which is the record's DD95 plus four points; do not change settings during a drawdown; review monthly. In the fourth month the account is down 13 per cent and the trader wants to stop. The plan says 22. They continue. The drawdown reaches 16 per cent in month five and recovers by month eight. The trader who stopped at 13 took a 13 per cent loss on a strategy that went on to recover, and that is the most common way a working EA loses money for its owner.
An EA is consistent. Its owner is the variable. Every intervention, raising risk after a good run, cutting it after a bad one, switching off in a drawdown, restarting after the recovery, changes the result the EA would have produced, almost always for the worse. The plan exists to make the owner as consistent as the program.