Building a system you can actually test

Ask most struggling traders what their strategy is and you get a description of a feeling: "I look for momentum near support and get in when it looks strong." That cannot be tested, cannot be reviewed after a losing run, and cannot be improved — because there is nothing fixed to compare against.

A system is testable when someone else could follow it and reach the same decisions you would. That is a high bar, and it is the whole bar.

The five rules

Write these down before you trade the idea. On paper or in a document, in specific language.

1. Entry — what must be true

Not "when it looks strong". Something checkable: which pair, which timeframe, what condition, at what time of day. "EUR/USD on the 1-hour chart, when price closes above the previous day's high during the London session" is testable. Whether it works is a separate question — but you can now ask it.

2. Exit — where you take the win

Defined before entry, because afterwards you will be negotiating with yourself. A fixed target, a trailing rule, a time-based exit — any of them work. "When I feel it's run out of steam" does not, and it is why so many traders hold winners until they become losers.

3. Stop — where the idea is wrong

A price level that disproves your reason for the trade, placed on the chart rather than derived from your tolerance for loss. This is covered in full in stop losses that mean something.

4. Size — what being wrong costs

A fixed fraction of the account per trade, converted into a lot size by the stop distance. Fix the percentage, not the lots: 1% of the account, whatever that means in lots today. The method is in position sizing, and the calculator does the arithmetic.

5. Conditions where it does not apply

The rule almost nobody writes, and the one that saves the most money. Every strategy has an environment it was built for. A breakout system fails in a range; a mean-reversion system is destroyed by a trend. State the conditions under which you sit out — around major scheduled news, in the thin hours between the New York close and the Tokyo open, when spread is unusually wide.

Without this rule you will trade the system everywhere, discover it fails in some conditions, and conclude the system is broken when it was only being used outside its range.

Making it survive contact with a real account

Two constraints belong in the written plan, not in your memory.

Leverage is a ceiling, not a target. Retail clients are capped at 30:1 on major pairs (3.33% margin), and your broker must close positions when funds fall to 50% of required margin. If a normal losing streak at your chosen size brings you near that line, the size is wrong — the rule exists to stop catastrophe, not to be planned around.

A daily or weekly loss limit. "After three losing trades, or 3% down in a day, I stop for the day." This is the only rule that reliably interrupts revenge trading, because it does not require you to be calm — it requires you to have decided in advance.

Then test it, and expect it to fail

With five rules written, you can finally run a fair test — and most systems fail one. That is a result, not a setback: rejecting an idea on paper costs nothing, while discovering it live costs the account.

Test the rules exactly as written, count every variant you try, and hold back data you have not looked at. The methods, and the reason a heavily tuned backtest means so little, are in backtesting honestly.

One warning worth carrying: past performance does not indicate future results. The FCA requires firms to say so prominently whenever they show it, because historical results persuade people far more than they predict. Your own results are subject to exactly the same limitation, and you have a much stronger motive to believe them.

The plan is the deliverable

The written plan is more valuable than any individual rule in it. It gives you something to review after a bad week that is not your memory — which will have quietly rewritten what you intended. Traders who keep one improve; traders who keep the plan in their head repeat.