How to paper trade properly, so that it proves something

What paper trading can and cannot tell you

It cannot tell you whether the strategy will be profitable. Nobody can tell you that, and two weeks of simulated gains is not evidence.

It can tell you four concrete things: whether your signals fire when the backtest said they would, whether the fills you get resemble the fills you assumed, whether your costs match your model, and whether you personally can execute the rules without improvising. Those four are worth a great deal and none of them require the strategy to be profitable.

Set it up like a real account

• Use the same account size you intend to fund. A demo account with a hundred times the capital produces different position sizes and therefore different behaviour.
• Use the same broker and the same symbol, so the spread you experience is the spread you modelled.
• Apply the same position sizing rules, including the rounding down to the minimum lot.
• Write down the rules before you start, and treat any deviation as a recorded exception rather than an adjustment.
• Run it for a duration defined by trade count, not by calendar days.

Record the comparison, not just the result

The most valuable output of paper trading is a table comparing what the backtest expected with what actually happened. Build it trade by trade.

The duration question

Two weeks is meaningless for a strategy that trades twice a month, and far too short even for one that trades daily. The right measure is the number of trades observed relative to the number in the backtest's typical run.

A practical target: keep going until the paper run has produced at least as many trades as the backtest produced in its worst quarter. That way the paper period includes something resembling a difficult stretch, not just a favourable one.

The part nobody wants to hear

Paper trading removes the emotional cost of loss, and that cost is a real part of live performance. A strategy that you execute perfectly with fake money may be one you abandon in week three with real money.

This is not a reason to skip paper trading. It is a reason not to treat a clean paper run as proof of anything except that the mechanics work. The behavioural test only happens when the money is real, which is an argument for starting with an amount small enough that the answer does not matter.

When to stop paper trading

Stop when the comparisons above show no unexplained mismatches and the trade count is adequate. Do not stop because the equity curve looks good, and do not continue indefinitely as a way of avoiding a decision.

The goal is not confidence. It is to have replaced assumptions with observations, so that whatever happens next, you know which parts of your model of the world have been checked.