Paper trading mistakes that make results misleading
Practice results are easy to flatter without noticing. These are the specific habits that make a paper trading record look better than the decisions behind it — and what each one costs you when the money becomes real.
Logging the trade after you know how it ended
The most common and most damaging one. If you write the entry once the outcome is known, your recorded reasoning is a reconstruction, and reconstructions are always tidier and more decisive than the original thought.
What it costs: your journal becomes a record of a competent trader who does not exist, and the review teaches you nothing because there is nothing wrong in it.
Fix: write the intent before the entry, even one line. Log the outcome separately.
Leaving out costs
Fees, spreads and, for anything held overnight with leverage, financing. A strategy that nets a small edge per trade can be entirely consumed by costs, and a simulated record that ignores them will never show you that.
What it costs: high-frequency approaches look viable in practice and lose money live, because costs scale with the number of trades.
Fix: subtract realistic costs from every logged result, including the losers.
Assuming a perfect fill
A simulation fills you at the price you name. Real markets do not, particularly when you most want out. Gaps, widening spreads and thin books are not edge cases; they cluster exactly around the volatile moments your strategy probably depends on.
What it costs: the worst real trades are systematically worse than the worst practice ones, which is the opposite of the direction you can afford to be wrong in.
Fix: assume a worse fill than you would like, especially on exits during fast moves.
Sizing bigger than you ever would
A simulated balance costs nothing, so it is tempting to trade a size that produces interesting numbers.
What it costs: you rehearse a version of position sizing you will never use, and the emotional rehearsal is wrong too — a 40% simulated drawdown is a shrug; the real one is not.
Fix: size as if it were your own money, because the point is to practise the decision.
Quietly not logging the bad ones
Nobody decides to do this. It happens by attrition: the trade was impulsive, you already feel foolish, and logging it makes it permanent.
What it costs: your win rate and profit factor describe a filtered sample, and filtered in precisely the direction that hides the behaviour you need to see.
Fix: treat the log as complete or treat it as worthless. A calendar makes gaps visible, which is uncomfortable and useful.
Moving the exit after entering
Deciding where you are wrong, then deciding you are not wrong yet. On paper this often works, because positions that would have stopped you out sometimes recover.
What it costs: it teaches you that holding through your own invalidation is rewarded. It is rewarded until the one time it is not, and that time is usually large enough to matter.
Fix: log stop moves as their own event. If the checklist shows "moved my stop" on a quarter of your trades, that is the finding.
Judging decisions by outcome
Marking every winner as a good trade and every loser as a mistake.
What it costs: in any process with a random component, outcome-grading trains you toward whatever happened to work in a small sample. You will drop correct trades that lost and repeat reckless ones that won.
Fix: score the process separately from the money. RATTLE's checklist is deliberately scored from what you ticked, not from the P&L.
Concluding too early
Fifteen trades, profit factor of 2.4, and a decision that the approach is validated.
What it costs: small samples swing violently on single results. A couple of ordinary winners can carry an unremarkable approach to impressive-looking figures, and the correction arrives with real money attached.
Fix: treat early metrics as description, not evidence, and say the sample size out loud whenever you quote one.
The honest summary. Every item above makes a practice record look better than the trading behind it. None of them is dishonest in intent; all of them are easy. If your simulated results seem strong, the most useful question is which of these is doing the work.
Next
If you are starting out, the ten-trade starting routine is built to avoid most of these by default. If you are weighing practice against historical testing, see paper trading versus backtesting.