What it means
Log every setup, hide the right side of the chart, include costs, and keep some data back for an out-of-sample check.
In practice
A backtest answers the first question about any idea: would these rules have made money? Traders go through past charts one candle at a time, record every setup that meets the rules, and work out the win rate, average result and worst drawdown.
Beyond the numbers, it builds familiarity with how the setup looks and behaves, which makes it easier to follow in real time.
What to watch for
It is easy to cheat without meaning to: skipping ugly setups, entering on information that came later, or leaving out costs. Good past results are a minimum requirement, not a promise about the future.
Related terms
4 terms- OverfittingTweaking rules until they fit the past perfectly, so they fail on new data.
- Look-ahead biasUsing information you couldn't have had at the time, such as a candle that hadn't closed yet.
- Forward testingTrading a strategy in real time, first on demo and then with very small size, before scaling up.
- Sample sizeHow many trades your conclusions rest on; under about 30 is mostly noise, aim for 50 to 100 or more.