How to test a trading claim

In one line: Turn a guru's sentence into a fair test with costs.

1. Write it down exactly. 'It works' is not testable. 'Buy when X, sell when Y, stop at Z' is.

2. Pick data before you look at results. Keep part of it aside: In-sample vs out-of-sample.

3. Include costs: Fees and commission, Bid, ask and spread, Slippage.

4. Compare with chance: Random entry (random twin). If coin flips do as well, the rule adds nothing.

5. Count enough trades: Sample size. And don't try 100 versions and keep the best: P-hacking and data snooping.

Our full method is on How we test.

Sources & studies

GuidesBacktestRandom entry (random twin)

Last checked: 2026-11-04 · Educational, not financial advice.

Education only, not financial advice. No signals. Backtests are hypothetical: real trading can be worse. Past results don't predict future results.