How many trades does it take to prove a strategy?
Enter a strategy's average return per trade and how much single trades vary. The calculator shows how many trades, and how many months, it takes before the result means something.
Calculations run in your browser. Nothing you type is sent or stored. Results are estimates for information only, not investment advice.
How it works
To tell an average return μ apart from zero when single trades swing by σ, you need about n = ((zconfidence + zpower) × σ ÷ μ)² trades. With 90% confidence and 90% power both z values are 1.28. Confidence is how sure you want to be that a positive result is not luck; power is how likely the test is to notice an edge that is really there.
The t-value after one year is μ ÷ σ × √(trades in a year). Values around 2 or more are the usual bar for taking a result seriously; below 1 the year tells you almost nothing. Real trades are not independent, and markets change, so treat the answer as a lower bound.
Questions
How long did your own strategy need?
Our daily and 4-hour trend ensemble averaged +3.7% per trade with a standard deviation of 18.7% and about five trades a month: roughly 167 trades, or 34 months, at 90% confidence and power. The full calculation.
Can a two-week live test prove anything?
Not about profitability. In nine years of replay, 60% of all two-week windows of our breakout rule ended at or below zero. Short tests can check that the system follows its rules, not that the rules make money. Our two-week test.