How far back should you backtest a system for you to have confidence in it?
The longer the time period, the better. That means more good data points to test against. You want to see how the system works during different market cycles, trending and not, and volatility levels. Systems work until they don’t. If you figure out when they do and don’t work, more likelihood of success.
I’ve know traders to go back several months, which I don’t think is far enough. I’d go several years.
And then forward test it as well.
Hi 
I using data at least 2008. Regards Greg
As long as possible. Will only work in your favour, not against. And will give you that much more confidence.
I have gone for as low as 6 months and as high as 2 years but there are some traders who perform it for over 5 years.
As far possibly that you can go back, the better it is.