Posted equity curves and what is not in the screenshot
Posted: Sat Oct 31, 2026 8:23 am
Every equity curve on a forum is a survivor. Ten people try variants of an idea, and the one with the smoothest curve posts it. Nobody lies, it is just selection. A screenshot labelled backtest carries no information about how many versions were tried before that one.
Before copying anything, a few questions are worth asking:
- how many parameter sets were tested to find it
- which spread, commission and slippage assumptions were used
- whether the test period includes the data used to build the rules
- whether there is a broker-verified live statement, not just a tester report
A crude mental discount helps: the more attempts behind a result, the less it means. Pick the best of fifty random systems and a strong-looking curve is almost guaranteed.
The same applies to personal results. The system that got kept is the survivor of everything tried. Keeping a plain log of every variant tested, failures included, gives a real trial count and makes it easier to be sceptical of the winner.
Do you keep a log of failed variants, and how many attempts sat behind the last system that looked good?
Before copying anything, a few questions are worth asking:
- how many parameter sets were tested to find it
- which spread, commission and slippage assumptions were used
- whether the test period includes the data used to build the rules
- whether there is a broker-verified live statement, not just a tester report
A crude mental discount helps: the more attempts behind a result, the less it means. Pick the best of fifty random systems and a strong-looking curve is almost guaranteed.
The same applies to personal results. The system that got kept is the survivor of everything tried. Keeping a plain log of every variant tested, failures included, gives a real trial count and makes it easier to be sceptical of the winner.
Do you keep a log of failed variants, and how many attempts sat behind the last system that looked good?