How many trades before a scalping edge is believable?
Posted: Sat Oct 31, 2026 8:21 am
Quick maths. A 58 percent win rate over 100 trades has a standard error of about 4.9 points, since the square root of 0.58 times 0.42 divided by 100 is roughly 0.049. So the true win rate could plausibly sit anywhere from the high 40s to the high 60s. With a 1:1 payoff, that range covers losing money and printing it. At 1,000 trades the standard error shrinks to about 1.6 points, which is far more useful.
But trade count is not the only thing that counts. A thousand trades taken in one calm month under one regime are worth much less than a thousand spread across trending weeks, ranging weeks and a couple of news-heavy periods. The trades are not independent samples of the future.
Look at average pips per trade with its own standard error, not only the win rate, because payoff size varies. If the average is +0.3 pip and the error bar is 0.4, nothing has been proven yet.
A reasonable personal rule: no size increase until several hundred trades covering at least two different conditions, with the confidence band clearly above the cost estimate.
What is your minimum trade count before increasing size, and how did you arrive at it?
But trade count is not the only thing that counts. A thousand trades taken in one calm month under one regime are worth much less than a thousand spread across trending weeks, ranging weeks and a couple of news-heavy periods. The trades are not independent samples of the future.
Look at average pips per trade with its own standard error, not only the win rate, because payoff size varies. If the average is +0.3 pip and the error bar is 0.4, nothing has been proven yet.
A reasonable personal rule: no size increase until several hundred trades covering at least two different conditions, with the confidence band clearly above the cost estimate.
What is your minimum trade count before increasing size, and how did you arrive at it?