Testing tells you what to expect, not what will happen
A backtest cannot tell you a system will make money. It can tell you what the system did over a defined period, which is the closest thing to evidence available before risking capital.
The gap between those two statements is where most backtesting goes wrong.
The two honest reasons to backtest
To find out whether an edge plausibly exists. Over 200 historical trades, was expectancy positive? If not, no amount of live conviction will fix it.
To learn your own numbers. Expected win rate, average win in R, average loss, worst losing streak, worst drawdown. This second reason is the more valuable one — because when the worst streak arrives live, the difference between meeting a forecast and meeting a surprise decides whether you keep executing.
The failure that invalidates most retail backtests
Hindsight. You are scrolling a chart where the outcome is already visible. Your eye finds setups that worked. You do not notice the ones you would have taken at the time and skipped in hindsight.
This is not carelessness — it is how visual pattern-matching works, and it is nearly impossible to avoid by trying harder.